Blog

Claude Fable 5: The First Mythos-Class Model You Can Actually Call Every so often a model release shifts the ceiling rather than nudging it, and Claude Fable 5 is one of those. Announced by Anthropic on 9 June 2026, Fable 5 is the first Mythos-class model made generally available — a new tier that Anthropic positions above the Opus class in raw capability (Anthropic, “Claude Fable 5 and Claude Mythos 5”). If you have been building on Opus or Sonnet, most of your code will keep working. But Fable introduces one genuinely new behaviour — it can decline a request and hand it to a different model — and if you do not plan for that, your integration will do something surprising in production. So this post does two things: it explains what actually makes Fable different, and it gives you a working Python example that handles the new behaviour without falling over. Think of it like hiring a brilliant but heavily vetted new colleague. They are the sharpest person in the building, but on certain sensitive topics they are contractually obliged to pass the ticket to someone else. Useful to know before you route all your traffic to them.... Claude Fable 5: Anthropic's First Public Mythos-Class Model, and How to Build With It

Introduction: Explainable AI with SHAP and LIME in Python Black-box machine learning models can be difficult to trust and debug, especially in domains where decisions have significant consequences such as healthcare, finance, and criminal justice. A model that quietly denies someone a loan without anyone being able to say why is not just a technical problem — it is a governance problem. Explainable AI (XAI) addresses this challenge by providing methods and tools to understand how AI models make decisions. Two of the most powerful and widely used libraries for model interpretability are SHAP (SHapley Additive exPlanations) and LIME (Local Interpretable Model-agnostic Explanations). In this tutorial, I explore how to use SHAP and LIME in Python to make machine learning models transparent and interpretable. We will cover both global and local explanations, understand feature importance, and learn how to communicate model decisions effectively. Every code block below was executed and verified against shap 0.49.1, lime 0.2, scikit-learn 1.7.2, and xgboost 2.1.4 — worth mentioning because the SHAP API changed substantially around version 0.42, and a lot of older tutorials (including an earlier draft of this one) silently produce wrong results on current versions. I will point out those pitfalls as... Explainable AI with Python: SHAP and LIME for Model Transparency

Introduction This week the frontier stopped looking like a single race and started looking like a shelf. OpenAI shipped a three-model family in one GA push, SpaceXAI undercut Opus on price and token count, and Meta priced agentic capability for the first time, with comparatively little fanfare. Mistral kept giving away a formal-proof specialist and a robotics model under an open licence while the closed labs argued over tiers. Underneath it all, Sysdig’s JadePuffer was a reminder that the same autonomy everyone celebrates in a launch post works exactly as well without a human pointing it at something useful. I’ll take the frontier releases first, then the open-weight side, then security, governance, and research. In this issue: GPT-5.6 Arrives as OpenAI’s New Flagship Family Grok 4.5 Ships Publicly as an Opus-Class Coder Leanstral 1.5 — Mistral’s Apache 2.0 Proof Engineer JadePuffer — The First Fully Agentic Ransomware The UN’s First Global Dialogue on AI Governance ICML 2026 Opens in Seoul on a Record 23,918 Submissions Frontier Models 1. GPT-5.6 Arrives as OpenAI’s New Flagship Family GPT-5.6 — OpenAI, 9 July 2026 OpenAI rolls out GPT-5.6 — Engadget, 9 July 2026 OpenAI's gpt-realtime-2.1 and mini — MarkTechPost, 6 July 2026 OpenAI... AI Weekly Signals: The Frontier Splits Into Tiers

Introduction Four frontier models moved this week, and only some of them moved freely. Anthropic put Claude Sonnet 5 in front of every user at a price that undercuts its own previous generation, and Google made computer use a standard tool inside Gemini 3.5 Flash rather than a separate premium model. Meanwhile, OpenAI kept GPT-5.6 gated to roughly twenty government-vetted organisations, and reportedly opened a much bigger conversation: handing the US government a 5% stake in the company itself. The pattern from last week was capability versus access. This week, access itself began to carry a price tag. I will take them in the order they landed. In this issue: Claude Sonnet 5 — Anthropic’s New Default, Priced to Undercut Itself Gemini 3.5 Flash Gets Computer Use — Google Makes Agent Access the Default Grok 4.5 — Private Beta, Kept In-House at SpaceX and Tesla GPT-5.6 Sol, Terra, Luna — Previewed, Then Gated to ~20 Vetted Organisations OpenAI Reportedly Floats a 5% Government Stake Claude Fable 5 and Mythos 5 Return After 19 Days Offline Anthropic Launches Claude Science and an In-House Drug Discovery Program Frontier Models 1. Claude Sonnet 5 — Anthropic’s New Default, Priced to Undercut Itself Introducing... AI Weekly Signals: Tokenizer Tax, Cache Rules, and Who Owns AI's Upside

What Is GEO (Generative Engine Optimization)? A reader asked me a question the other week, then told me they had already “checked with ChatGPT” first. The AI gave them a decent answer — and cited someone else’s blog. I had written about the exact topic. My post just wasn’t structured in a way the AI could lift. That is the whole problem GEO solves. Generative Engine Optimization (GEO) is the practice of structuring content so AI answer engines — ChatGPT, Perplexity, Google AI Overviews — can parse it, quote it, and cite you as a primary source. In this post, I explain what GEO is, how it differs from SEO, and exactly what I changed across roughly 95 posts on this blog, including the first results. GEO vs SEO: What Actually Changed Search Engine Optimization (SEO) optimises a page to rank in a list of links that a human then clicks. GEO optimises the same page to be extracted and repeated inside an answer the AI writes for the user, often with no click at all. They are not enemies. The same crawlable, trustworthy, well-written page feeds both. But the unit of success is different, and that changes what you... What Is GEO? How I Optimised My Blog for AI Search

Introduction Last week the open-weights story hinted that the model itself is becoming a commodity. This week proved it by omission: almost nothing that mattered was a model release. The contest has quietly moved to every layer wrapped around the model — the silicon underneath, the capital behind the silicon, the orchestration bolted on top, the security boundary nobody was guarding, the regulators circling the cashier, and the human still sitting in the chair. If you only watch the leaderboard, this was a dull week. If you watch where the leverage actually sits, it was one of the busier ones I have tracked. I will take them in the order they landed. In this issue: Qualcomm Circles Tenstorrent — RISC-V Takes a Run at Nvidia China Drafts a $295bn AI Build-Out — Without Nvidia OpenRouter Fusion — A Panel of Cheap Models Beats the Frontier Agentjacking — Your Coding Agent Is the Attack Surface 42 State Attorneys General Subpoena OpenAI Anthropic: 400k Sessions Say Domain Expertise Beats Coding Skill Silicon and Compute: The Nvidia Monoculture Under Pressure 1. Qualcomm Circles Tenstorrent — RISC-V Takes a Run at Nvidia Qualcomm mulls taking over Jim Keller's Tenstorrent — deal would value the... Qualcomm, RISC-V, and the Crack in Nvidia's Monopoly

Fixing Docker “Permission denied” for a Non-Root Container User A Docker permission error is an ownership mismatch: a folder created by root (at image build or container startup) is being written to by an application that runs as a non-root user such as appuser, so Linux refuses the write. The fix is to change folder ownership with chown, not to loosen permissions with chmod 777. Sometimes Docker does not fail loudly. It does not say: Hello Elena, your container is fine, but your app cannot write to the folder because the owner is wrong. No. Docker prefers mystery. You run the app. Everything looks healthy. The container is up. Nginx is serving. Flask is alive. Then suddenly uploads fail, backups do not appear, static files refuse to update, or Python throws one of those wonderfully unhelpful messages: Permission denied In my case, the fix was this command: docker compose exec -u root web chown -R appuser:appuser /backups /app/protected_files /app/app/static At first glance, it looks like a scary Linux spell. But it is actually a very practical Docker permissions repair. Let’s unpack it. The short version This command changes the owner of three important folders inside the running web container: /backups... Docker Permissions Without Panic: Why I Ran chown Inside My Container

Introduction layout: post title: “Git Origin Already Exists: How to Add, Delete, and Manage Git Remotes” subtitle: “A practical guide to Git origins, remote URLs, and fixing the classic ‘remote origin already exists’ error.” description: “Learn what Git origin means, why the ‘remote origin already exists’ error happens, and how to add, remove, rename, and update Git remotes safely.” date: 2026-06-20 categories: [Git, Developer Tools] tags: [git, github, version control, command line, developer workflow] ———————————————————————- You create a new GitHub repository. You copy the helpful command from GitHub: git remote add origin git@github.com:your-name/your-repo.git And Git answers, rather firmly: error: remote origin already exists. At first glance, this feels like Git being difficult for no reason. But this error is actually useful. It means your local repository already has a remote called origin. Git is not refusing to work. It is telling you: “I already have a remote with that name. Please decide whether you want to use it, change it, rename it, or remove it.” Let’s unpack what origin means, why this error happens, and how to manage Git remotes without turning your repository into a small haunted forest. What is origin in Git? In Git, a remote is a... Git Origin Already Exists: What It Means and How to Fix It

The previous post covered building a private AI lab: Ollama, Open WebUI, and a RAG pipeline that stays entirely on your hardware. That stack is straightforward to secure — it reads and responds, and the threat model is simple. This post is about the next step: autonomous agents. Tools like OpenClaw (also known as Molt or Moltbot) and Aider can do things on your behalf — read your email, write and run code, manage files, call APIs. That capability changes the security picture in ways that are worth understanding before you enable it. The core problem has a name from computer security: the confused deputy problem. What is a confused deputy? The “confused deputy” is a classic concept from operating system security. A deputy is a program that has permissions to do things on your behalf. The confusion happens when the deputy receives instructions from a source other than you — and cannot tell the difference. For AI agents, the deputy is the agent itself. You give it access to your email, your file system, your APIs. It has your permission to use those. The confusion happens when it reads a document, a web page, or an email that contains... Agents, Access, and the Confused Deputy Problem

Open Weights, AI Debt, and Model Access: The Week in Context A quieter week for raw model launches, and a much louder one for the money and machinery underneath them. The headline release was open-weight, the headline numbers were denominated in billions of dollars of debt, and the most capable model from last week was still switched off by government order. If you only track leaderboards, you missed most of what mattered. The thread running through it is ownership: who owns the weights, who owns the compute, who owns the debt that paid for the compute, and who gets to decide — on no notice — that a model you depend on is no longer available. Capability was almost a side plot this week. I will take them in the order they landed. In this issue: Z.ai’s GLM-5.2 — Open Weights Catch the Frontier SpaceX Lines Up a $20bn Bond Sale Google Builds a TPU Business — and Rents It to Anthropic Elastic Buys Deductive AI to Clean Up After the Robots Fable 5 Stays Offline — and “Hardware Sovereignty” Goes Mainstream Open-Weight Models and Inference: GLM-5.2 vs the Closed Frontier 1. Z.ai’s GLM-5.2 — Open Weights Catch the Frontier... Open Weights, Big Debt

Introduction I expected Claude Code to be another AI coding assistant with a nicer interface. You know the type: autocomplete with opinions, a chat window that occasionally suggests a useful function, and a great deal of confidence around edge cases it has not actually tested. Claude Code feels different because it does not stay politely outside the project. It enters the repo. It reads files, edits code, runs commands, checks errors, creates commits, and shows you the diff. Used without caution, that sounds alarming. Used carefully, it has become one of the most useful workflow tools I have added this year. The interesting part is not that Claude can write code — we already knew models could do that. The interesting part is where it sits: inside the terminal, the editor, the Desktop app, the git history, and all the small maintenance tasks that normally eat an afternoon. This post is a practical field note: where I reach for Claude Code first, where I deliberately slow it down, and how I use it both as a developer and as someone maintaining a technical blog. The Shift: From Prompting to Delegating The biggest change for me was not learning a new... How I Actually Use Claude Code: CLI, Desktop, Diffs, and Blog Workflows + Free Guest Passes inside!

AI Governance Convergence: Federal Framework, IPO Filings, and Frontier Model Access This was the week AI governance stopped feeling theoretical. Congress published a 269-page federal AI framework. OpenAI confirmed a confidential S-1. Anthropic released Fable 5 while keeping its most capable model behind a restricted access layer. Apple opened Foundation Models to small developers at no cloud API cost. Jeff Bezos brought a $41 billion physical-world AI company out of stealth. And Oracle quietly removed one of the biggest friction points in enterprise AI procurement. The common thread is not another benchmark race. It is control: who funds frontier AI, who audits it, who gets access to the strongest models, and on what terms. In this issue: Great American AI Act: Congress Tables Its 269-Page Framework OpenAI Files Its S-1 at $852 Billion OpenAI + Oracle Cloud: Enterprise Distribution at Scale Bezos’ Prometheus Exits Stealth at $41 Billion Apple WWDC 2026: Foundation Models Go Free for Small Developers Claude Fable 5 and Mythos 5: Anthropic’s Strongest Model, with Caveats Governance & Policy 1. Great American AI Act: Congress Tables Its 269-Page Framework Obernolte, Trahan Release Discussion Draft of the Great American AI Act — Representative Jay Obernolte, 4 June 2026... Who Controls AI? Fable 5, OpenAI's S-1, and a 269-Page Bill

I noticed something odd while working with AI coding agents: I was not only prompting them. I was slowly teaching them my habits. How I review Python snippets. How I prepare a newsletter. How I want blog posts to sound. How I prefer Markdown to be preserved. How do I check that an AI has not enthusiastically “improved” something that was already fine? At some point, repeating those instructions stopped feeling like prompting and started feeling like maintaining tiny workflow manuals. That is where AI skills come in. AI Skills Defined: Reusable Workflow Packages for AI Assistants and Agents AI skills are reusable instruction packages for AI assistants and agents. Think of AI skills as a small “how we do this task” folder. Instead of pasting the same long prompt every time, you turn the workflow into a skill once, then the AI can reuse it when the task comes up. A skill usually contains: my-blog-newsletter-skill/ ├── SKILL.md # main instructions ├── examples/ # good examples ├── templates/ # output templates ├── scripts/ # optional helper code └── references/ # style guide, rules, schemas, checklists The important file is usually SKILL.md. It explains what the skill does, when to use... How AI Skills Turn Workflows into Reusable Instructions

AI Weekly Signal: Claude Opus 4.8 Dynamic Workflows, Open-Weight Model Releases, and GitHub Copilot Metered Billing Some weeks in AI feel like steady background noise. This week — was not one of those weeks. Claude Opus 4.8 shipped dynamic workflows for parallel agentic systems. MiniMax M3 arrived as the first open-weight model combining a 1M-token context window with frontier-tier coding. Google released Gemma 4 12B, a multimodal model that runs agentic workflows locally on 16GB of RAM. Microsoft unveiled seven homegrown MAI models at Build 2026. GitHub Copilot switched to AI Credits metered billing. Anthropic filed a confidential S-1 with the SEC. And OpenAI embedded GPT-Rosalind into US biodefence infrastructure. The running theme, if there is one, is consolidation: the big players are locking in their positions, the cost of intelligence is being priced in real time, and the gap between “AI research preview” and “regulated sovereign infrastructure” is closing faster than most practitioners expected. 1. Claude Opus 4.8: Agentic Work Gets Serious Released: 28 May 2026 Source: Anthropic, TechCrunch Dynamic workflows is a multi-agent orchestration pattern in Claude Code that enables a single orchestrator session to spawn hundreds of parallel subagents, each with its own context window, then aggregate... GitHub Copilot AI Credits & Claude Opus 4.8

Local AI Agent Architecture with Cline, Ollama, and Model Context Protocol There is a particular kind of developer frustration that arrives by surprise, usually around the third subscription confirmation email of the week. One tool writes code, another reads your files, another connects to GitHub, another queries your database — and each one wants API credits, a seat licence, or a monthly plan. At least one of them is happily burning tokens in the background while you are still deciding what to type. At some point, the question is no longer “can this AI assistant help me?” It becomes “can I actually understand what it is doing, what it can reach, and what it will cost me?” That is what drew me to the combination of Cline + Ollama + MCP. Not because it is perfect — it is not — but because each part has a clear, honest role, and together they produce a local AI coding setup where the pricing story does not require a magnifying glass and a lawyer. Cline is an open-source AI coding agent for VS Code that connects local LLMs to MCP tool servers, requiring explicit user approval before executing any terminal command or... Local AI Agents with Cline, Ollama, and MCP

AI Weekly Signal: OpenAI IPO, DeepSWE Benchmark Gap, Glasswing Vulnerabilities, and Illinois SB 315 A quieter week for model releases — no major frontier launch from any of the large labs. What it delivered instead was seven stories that map the territory around the models: who funds them, who governs them, how to benchmark them honestly, what they can do to critical infrastructure, and whether hardware sovereignty is becoming real rather than aspirational. I will take them in the order they landed. In this issue: OpenAI and Anthropic move toward public-market scrutiny Pope Leo XIV frames AI as a question of human dignity GLM-5.1 pushes inference speed on Huawei chips Anthropic’s Project Glasswing shifts the security bottleneck to patching Illinois passes third-party audit rules for frontier AI DeepSWE exposes a wider gap between coding models Cognition’s $26B valuation shows how investors price AI coding agents Frontier AI Capital Markets: OpenAI and Anthropic IPO Filings in 2026 1. OpenAI moves toward a trillion-dollar IPO The big questions OpenAI's trillion-dollar IPO filing may finally answer — Fortune, 22 May 2026 OpenAI IPO 2026: What the Confidential Filing Means — Nerd Level Tech, 22 May 2026 US funds set aside cash as SpaceX... Claude Haiku scored zero. GPT-5.5 scored 70%. The new benchmark explains why

Introduction I have been slowly automating the repetitive parts of running this blog. Not the writing ideas — that stays human. But the publishing chores: resizing images, preparing metadata, checking links. The latest experiment was a Pinterest pin generator. The script reads a Jekyll post from _posts/, sends a cleaned-up version to Claude, and receives back a pin title, description, hashtags, and a short image excerpt. It then renders a 1000×1500 JPEG and uploads it to Pinterest via the v5 API. Before I ran this over my whole archive, I had one very practical question: how much will this actually cost? There is a specific kind of programmer’s dread that kicks in when you are about to loop a script over your entire post archive — not “will it work?” but “will it charge me forty dollars while I’m not watching?” So I decided to calculate it properly: not a rough guess, but an exact breakdown of what the script sends, how many tokens that produces, and what the API charges for it. This post walks through that calculation. If you are building something similar, the methodology is reusable for any prompt-based workflow. What the script sends The script calls... I used Claude to generate Pinterest pins. Here is the actual API cost.

Introduction This week had one unmissable centrepiece and three stories that reframe what it means. Google I/O ran on 19 and 20 May, and it was the most AI-dense developer conference Google has run. The keynote alone covered two new model families, a full agent development platform, a 24/7 personal AI agent, the largest redesign of Google Search in 25 years, and a new $100/month consumer tier. Google now processes over 3.2 quadrillion tokens per month — up seven times year-on-year. The Gemini app has 900 million monthly active users. These are not speculative numbers from a roadmap. They are the current state of the largest AI deployment on the planet. On the same day — 19 May — Anthropic ran Code with Claude London. The scheduling was not accidental. The night before, a jury in Oakland had taken 90 minutes to throw out Elon Musk’s $150 billion lawsuit against OpenAI. Google threw down the gauntlet, Anthropic fought for developer mindshare on the exact same day, and OpenAI had just dodged a $150 billion bullet. Forty-eight hours of high-stakes collision. I have picked four signals. The first is long — Google I/O deserves it. The other three are tighter. Models... Google I/O 2026: 900M Users, Musk Loses, Gemma Ships

Introduction The free tier has a ceiling, and it breaks in a specific order. First the message limit stops you mid-session — you are halfway through editing a document and Claude tells you it is done for the next few hours. Then, over weeks of use, the absence of persistent context starts to cost you time quietly: every session begins cold, every style guide gets re-pasted, every codebase gets re-explained. If you write code, the missing terminal tool is a different kind of gap — not an interruption but a capability you never had. Deep research is the last thing most people notice, and then only when a task actually demands it. Pro fixes all four, in roughly that order of urgency. At $20 a month ($17 billed annually) it is not cheap, but it is also not a minor incremental upgrade. This post covers each limit, what breaks when you hit it, and what the upgrade actually changes — without a conclusion about whether it is worth it. That depends on which of these you are hitting. THE FREE TIER CEILING — IN ORDER OF IMPACT FREE — WHAT BREAKS PRO — WHAT IT FIXES → 1 Message limit... Claude free tier limits: what breaks first and what Pro fixes

Introduction This is the sixth and final post in our Python Basics series. We have covered functions, error handling, and the standard library. Together with the earlier posts on basic syntax and OOP, you now have a solid foundation in Python. Generators are the topic I have been looking forward to most. They are one of those features that, once you understand them, you start seeing everywhere — and you wonder how you ever managed without them. The idea is simple and beautiful: instead of computing an entire sequence up front and storing it in memory, you compute one value at a time, only when it is actually needed. Lazy birds, in other words. They do not arrive all at once. They come one by one, when conditions are right. What Is an Iterator? Before generators, let us understand iterators, because a generator is a particular kind of iterator. An iterator is any object that implements two methods: __iter__() (which returns the iterator itself) and __next__() (which returns the next value, or raises StopIteration when there are no more values). You use iterators constantly without realising it. Every for loop in Python works by calling __next__() on an iterator: birds... Python Generators and Iterators: Lazy Birds

Introduction In the previous post we learned to handle errors gracefully. Now let us talk about something that will save you enormous amounts of time: Python’s standard library. The standard library is the collection of modules that ships with every Python installation. It covers file system operations, dates and times, data serialisation, mathematics, networking, testing, compression, and much more. The first instinct of many beginners — and honestly of many experienced developers — is to reach for a third-party package. Often the right answer is already in the box. The Python community has a saying: batteries included. This post is about finding the batteries. Modules and import Before exploring specific modules, a quick note on how to use them. A module is a Python file — or a package of files — that you bring into your script with import: import math print(math.sqrt(144)) # 12.0 If you only need one thing from a module, from x import y keeps the namespace cleaner: from math import sqrt, pi print(sqrt(144)) # 12.0 print(pi) # 3.141592653589793 You can alias a module to a shorter name with as: import datetime as dt today = dt.date.today() That is all there is to it. The rest... Python's Standard Library: Your Built-in Toolkit

Introduction Last week, the strongest signal was AI becoming the default layer: chat defaults, realtime voice, Gemini developer tools, Copilot governance, model testing, labour law, and data-center costs. This week, I want to avoid telling the same story again. The signal is still infrastructure, but it has moved closer to deployment. The important announcements were not mainly about one model beating another model. They were about how AI gets installed inside companies, how agents are allowed to touch code and systems, how legal teams can connect models to professional tools, how workspaces are turning into agent hosts, how phones may turn agentic AI into an operating-system feature, how serving infrastructure has to bend around agent-shaped conversations, how the open-model frontier keeps widening, and how the Python community is organising around all of it. What matters this week Deployment is becoming a product category. OpenAI is building a services arm around forward-deployed engineers, not only APIs. Cyber AI is moving into controlled access tiers. Daybreak ships with a roster of named launch partners, and OpenAI documents how Codex is contained in production. Vertical agents are getting professional plumbing. Claude’s legal connectors, the rebuilt CoCounsel Legal, and Notion’s External Agents API show... OpenAI’s $4B Deployment Bet

Introduction My suitcase was on the bed, half-packed, when I decided to fix the blog navigation. This is, in hindsight, not the ideal time to undertake a significant UX overhaul. But the blog has grown considerably over the past year — the series on AI tools alone runs to twelve posts, the Python basics series to six — and the tag list at the top of every /blog/ and /tag/ page had quietly become a wall of text. Dozens of pills, no grouping, no search, no sense of priority. Perfectly functional if you already knew what you were looking for. Not very useful if you did not. I had a flight to catch. I also had two AI coding assistants I had been meaning to compare in a real task rather than a contrived benchmark. The navigation overhaul became the experiment. This post is an honest account of how that went: what each assistant did well, where each one frustrated me, and what the final result looks like. The suitcase is zipped now. The navigation looks great on mobile. What Needed Fixing Before getting into the assistants, it helps to understand the starting point. The existing system had a flat... Two AI Assistants, One Blog Navigation Overhaul

Introduction In my first Python post we covered variables, lists, dictionaries, and list comprehensions — the data and control flow that let you write a working script. In the OOP post we jumped all the way to classes. But there is an important stop in between, and that stop is functions. Functions are how you stop writing the same thing twice. They are the reason a 500-line program does not become a 5,000-line program. They are also the first step toward thinking about code as something you design rather than something you just write. Once functions feel natural, classes make much more sense — a class is largely just a collection of functions that share some data. We will keep our birds. They are patient, useful, and by now familiar. What Is a Function? A function is a named block of code you can call by name, pass data into, and get a result back from. In Python you define one with def: def greet_bird(name): print(f"Hello, {name}!") greet_bird("Eagle") greet_bird("Pigeon") Hello, Eagle! Hello, Pigeon! That is the whole idea. Write the logic once inside def, then call it as many times as you need. Without functions we would have to repeat... Python Functions: Writing Reusable Code

Python Error Handling: Why Exceptions Matter In the previous post we wrote functions — clean, reusable pieces of logic. But functions assume that the inputs they receive are sensible, the files they open exist, and the network they query is available. In the real world, none of these things are guaranteed. Users make typos. Files get deleted. APIs go down. Disks fill up. The question is not whether your program will encounter an error. It will. The question is whether it will crash with a cryptic message and lose all its work, or handle the situation gracefully and tell you — or the user — what actually went wrong. Python’s exception system is your answer. And our birds, as patient as they are, are about to misbehave. What Is an Exception in Python? A Python exception is an object that represents an error detected during execution. When Python encounters a problem it cannot resolve, it raises an exception that represents what went wrong. If nothing catches it, the program stops and prints a traceback. You have certainly seen this before: wing_spans = {"Eagle": 200, "Pigeon": 50} print(wing_spans["Albatross"]) KeyError: 'Albatross' The KeyError is an exception. Python raises it because you asked... Python Error Handling: When Birds Misbehave

Introduction Last week, I wrote about new models, security risks, and the scaling ceiling. This week feels different. It is less about the model as an object and more about AI becoming part of ordinary systems: chat defaults, APIs, coding tools, government evaluation, labour law, electricity markets, and local permitting. The model is still important. But increasingly, the signal is not only what the model can do. It is where the model is placed, who controls access to it, how developers build around it, and who pays when its costs leave the data centre. I have picked the key signals this week, plus a short follow-up from last week’s DeepSeek story. Models and Defaults 1. GPT-5.5 Instant becomes the default — and defaults are distribution power OpenAI releases GPT-5.5 Instant, a new default model for ChatGPT — TechCrunch, 5 May 2026 OpenAI claims ChatGPT's new default model hallucinates way less — The Verge, 5 May 2026 OpenAI makes default ChatGPT more personal — Axios, 5 May 2026 OpenAI released GPT-5.5 Instant on 5 May and made it the new default ChatGPT model, replacing GPT-5.3 Instant for everyday use. It is not the most dramatic kind of AI announcement. There was... AI’s New Defaults and Hidden Costs

Introduction There was a time when reading and writing separated those who could participate fully in public life from those who could not. It was not just about decoding words on a page — literacy unlocked access to contracts, news, correspondence, knowledge, and economic opportunity. Those without it were not less intelligent; they were simply excluded from systems built around the assumption that you had it. I think we are living through a similar shift right now, and AI is at the centre of it. This is not a dramatic claim. I am not saying AI will replace everyone or that the world ends for those who do not adapt. What I am saying is more practical: AI tools are being woven into work, communication, healthcare, education, and daily decision-making faster than most people realise, and the gap between those who know how to use them well and those who do not is widening. Knowing how to work with AI — critically, deliberately, and effectively — is becoming a foundational skill, much like reading once was. In this post, I want to explore what AI literacy actually means in practice: what skills it involves, why they matter for everyone, and... AI is the New Literacy

Introduction This week brought something I have not seen quite so clearly before. Capability and constraint arrive at the same time. On one side, new models keep getting better — more capable, more autonomous, more useful in real workflows that people actually care about. On the other hand, the risks and limits are becoming genuinely impossible to ignore: cybersecurity threats, government involvement, and the sheer cost of running these systems at scale. For a long time, AI progress felt mostly one-directional. This week made something clear: Progress is now happening in tension with its consequences. One incident captures that tension better than any benchmark: in a controlled security test, an advanced model reportedly chained multiple vulnerabilities, escaped containment, and then posted proof of its own exploit path online without being explicitly asked. That kind of unsolicited initiative is exactly why capability gains now trigger immediate governance and release constraints. That is not a bad thing. In my opinion, it is a more honest picture of where we are. What happened this week OpenAI expanded the availability of GPT-5.5. DeepSeek released V4 (Pro and Flash), pushing open-weights performance and price-efficiency in coding-heavy workloads. NVIDIA announced Nemotron 3 Nano Omni, an open... Capability Meets Constraint

Introduction If you have been using Cursor AI for a while, you might have noticed that the assistant is great at reading and writing code, but it can only work with what you give it. It cannot peek into your database, check your API documentation, or inspect live logs on its own. MCP (Model Context Protocol) servers solve exactly this problem — they act as a bridge connecting Cursor’s AI assistant to external systems and data sources. In this post, I walk through what MCP servers are, how to configure them in Cursor, and how to write a simple one from scratch in Python. Why MCP Matters Without MCP, a typical debugging session looks roughly like this: Check the code in your editor. Read the logs in a terminal. Query the database in a separate client. Look up the API schema in a browser tab. Jump back to the editor to make changes. All that context switching is tiring and slow. MCP brings that external information directly into the AI assistant, so you can stay in one place and ask questions that span all of those sources at once. More precisely, instead of manually explaining your project’s structure to the... Cursor AI with MCP tools

This is Part 4, the final post in the Codex CLI series. We focus on advanced execution: scaling from personal usage to reliable team workflows. Introduction Hello, dear friends. In the first three posts, we covered installation, safety controls, and practical day-to-day workflows. This final part is about operational maturity: how to use Codex CLI in automation, how to troubleshoot quickly, and how to create team norms that stay safe under pressure. If Part 3 was about getting high-quality output, Part 4 is about getting predictable output. 1) Non-Interactive Automation with codex exec The core advanced feature is non-interactive mode (official docs). Why codex exec matters Interactive sessions are excellent for exploratory work. But production engineering needs repeatability. codex exec lets you run explicit tasks in scripts, CI pipelines, or scheduled jobs. Simple example: codex exec "Summarize open TODO comments grouped by directory" Permission defaults and escalation path Per docs, codex exec runs in a read-only sandbox by default. That is the correct baseline. Escalate only when needed: # Allow edits in automation codex exec --full-auto "Update stale docs links and propose minimal fixes" # Explicitly select workspace sandbox in CI codex exec --full-auto --sandbox workspace-write "Run tests, fix one... Codex CLI Part 4: Advanced Operations, Troubleshooting, and Team Patterns

Introduction Honestly, this week was the most consequential stretch of AI news in several months — and it ended with a bang. Nine signals worth covering, landing across seven days. OpenAI closed the week by releasing GPT-5.5 on 23 April — retaking the publicly available frontier lead and, more importantly, explicitly repositioning itself as an agent runtime rather than a chat model. Two Chinese labs shipped frontier-quality models on the same day earlier in the week: one proprietary, one fully open-source, both competitive with Western frontier systems. Image generation gained the ability to reason. Google confirmed its Gemini engine will power Apple’s next Siri. An infrastructure deal between Amazon and Anthropic locked in compute at a scale that changes the reliability picture. New chips from Google Cloud separated training and inference silicon for the first time. And the Stanford AI Index documented a field that has simply outrun every institution meant to guide it. The competitive map did not just shift this week — it moved in several directions simultaneously, and the open-source chapter of that story is now settled. Kimi K2.6 at #4 on the global intelligence index, level with the three major Western labs, is not a benchmark... Has the open-source gap closed?

This is Part 3 of the Codex CLI series. In this post, we move from basic usage to production-grade workflows for writing and shipping code. Introduction Hello, dear friends. In Part 1 and Part 2, we covered installation, safety, and control fundamentals. This post is about execution: the exact workflows you can run when you need results, not demos. I will focus on two areas: editorial blogging workflows in a Jekyll-style repository Python engineering workflows (refactoring, typing, tests, debugging) The core idea is simple: Codex works best when you give it a bounded task, explicit constraints, and a verification gate. Before we begin, one important update: the modern command for switching runtime permissions in the CLI is /permissions, while /approvals remains available as an alias (slash commands docs). A Practical Operating Model For daily work, I recommend this loop: Define scope in one sentence. Set a permission level that matches risk. Ask for a plan before large edits. Review /diff before accepting changes. Run objective checks (tests, linters, builds). Commit only after evidence is clean. This may feel strict, but strictness is what turns a coding assistant into a reliable collaborator. If you are new to the CLI interface, review... Codex CLI Part 3: Practical Workflows for Blogging and Python Development

Introduction Hi! Hope you are having a good week. I just returned from vacation — rested, curious, and apparently right on time, because this week’s stories were too good to miss. Something shifted this week. It was not about which model scored highest. It was about who gets access, under what rules, and what happens when those rules are tested. Anthropic wrapped a stronger Opus in cyber guardrails. OpenAI handed more powerful tools to verified defenders and pushed agents into real execution environments. Microsoft drove image generation further down the cost curve. And central banks issued warnings — in public, seriously. The stack got stronger. The world got serious. That is the real story this week. Model Releases and Agent Infrastructure 1. Claude Opus 4.7 — better coding, stronger vision, tighter cyber guardrails Introducing Claude Opus 4.7 — Anthropic Anthropic rolls out Claude Opus 4.7 as Mythos stays under lock and key — CNBC Anthropic released Claude Opus 4.7 on April 16, 2026. Anthropic says the model is a notable improvement on Opus 4.6 in advanced software engineering, especially on difficult tasks, and that it handles complex, long-running work with more rigor and consistency. The model also has better vision,... Agents, Cyber Models, and the Safety Stack Tightening Up

Introduction This week was not about volume — it was about intent. Compared to previous weeks, the pace of AI announcements slowed. But instead of signaling a slowdown, it revealed something more important: direction. Across multiple signals, a consistent pattern is emerging: Model releases are becoming more selective Platforms are integrating more tightly Efficiency is becoming a core priority This is what a maturing technology looks like. Let me walk you through the signals. What happened this week Meta released a new AI model, Muse Spark. Microsoft expanded its in-house multimodal AI model stack. New research highlights efficiency and optimization as key innovation areas. The pace of major releases appears more selective compared to previous weeks. Model Releases and Strategy 1. Meta launches Muse Spark, its new AI model Meta unveils first AI model from superintelligence team On April 8, Meta introduced Muse Spark, a new AI model developed by its superintelligence team. Key aspects: Multimodal capabilities Integration into Meta’s ecosystem Continued investment in advanced AI systems Takeaway: The frontier model race continues — with increasingly targeted releases. Why this matters to you The shift is subtle but important: Fewer headline launches More targeted deployment Tighter product integration 2. Microsoft... AI Signals: Controlled Releases and Platform Integration

Introduction This week made one thing very clear: AI is no longer just about models. For the past two years, the conversation has been dominated by capability — which model is smarter, faster, cheaper. That still matters, but it is no longer the center of gravity. What we are seeing now is a shift across the entire stack: From chips → to models → to interfaces → to market dynamics And importantly, all of these layers are starting to move at the same time. That creates a different kind of momentum — and a different set of risks. Let me walk you through the signals that stood out. What happened this week Microsoft launched new multimodal foundation models. Anthropic confirmed a powerful new model but is not releasing it yet. A startup raised $60M to use AI for chip design. Companies are preparing AI-native devices like smart glasses and earbuds. A new poll shows rising AI adoption but declining trust. AI startup valuations continue to surge at early stages. Model Releases and Safety Strategy 1. Microsoft releases new multimodal foundation models Microsoft releases new AI models to expand beyond OpenAI In early April, Microsoft introduced a new set of in-house... AI Signals: From Models to the Full Stack

The AI Paradox: Useful and Risky at the Same Time Modern AI agents do more than generate text. They read inboxes, browse docs, call APIs, run shell commands, and trigger workflows. That makes them useful. It also means a single hidden instruction in untrusted content can turn routine automation into a privacy or security incident. In this post, “persistent agents” means AI systems that keep memory or state across tasks and can repeatedly access tools, files, APIs, or workflows with limited human intervention. This is not an argument against agentic systems. It is an argument against giving them broad, persistent access without strong boundaries, narrow permissions, and reliable review paths. The core problem is not AI in the abstract. It is orchestration, permissions, and trust boundaries. If an agent can read untrusted content and call high-impact tools, your privacy and security posture depends on system design, not model quality alone. A Practical Threat Model for Persistent Agents Most avoidable failures follow the same chain: The agent ingests untrusted content. The model interprets part of that content as instruction rather than data. The planner or router selects a privileged tool. The tool executes before policy or human review stops it. A... The Digital Butler or Trojan Horse? A Privacy Playbook for Persistent AI Agents

Introduction This week felt like two very different AI stories happening at the same time. On one track, we got concrete, practical model releases — real-time voice and AI-generated music from Google. On the other, the constraints became more visible: energy and infrastructure pressure, data-privacy defaults, and a high-capability model leak that showed just how carefully labs are thinking about staged rollouts. I find this fascinating. For a long time, the only question that seemed to matter was: how capable is the model? Now, three equally important questions run alongside it: Can we power it? Are we allowed to deploy it? And who gets access first? This week illustrated all three constraints at once. Let me walk you through what happened. What happened this week U.S. lawmakers proposed a federal pause on new AI datacenter construction. GitHub changed how it uses Copilot interaction data for Free, Pro, and Pro+ users. AWS made Amazon Bedrock available in New Zealand for the first time. Google launched Gemini 3.1 Flash Live, a low-latency real-time multimodal model. Google launched Lyria 3 Pro, an extended music generation model, in public preview. Details about Anthropic’s unreleased Mythos/Capybara model leaked, and Anthropic confirmed it exists. Infrastructure and... AI's New Bottleneck

Introduction This week felt less like watching a model race and more like watching the foundations of a new industry being poured. While attention stayed fixed on the next benchmark or chatbot launch, the bigger story was happening lower down the stack. Nvidia used GTC to expand its hardware roadmap and push a broader Physical AI platform for robotics. Anthropic invested heavily in enterprise distribution and then showed an early version of asynchronous personal AI delegation. Mistral, OpenAI, and Microsoft all shipped notable updates in the efficiency tier within days of each other. And outside the usual US-centred spotlight, Xiaomi and Rakuten offered two different signs that the open-weight race is becoming both global and politically messy. What matters this week Nvidia pushed agentic AI and robotics as infrastructure problems, not just model problems. Anthropic signalled that enterprise distribution is becoming a moat. Dispatch hinted at a shift from synchronous prompting to asynchronous AI delegation. Mistral, OpenAI, and Microsoft all pushed the efficiency tier forward. Xiaomi and Rakuten showed that the open-weight race is now global and increasingly messy. Together, these signals point in the same direction. Value is migrating away from raw model capability and toward who controls the... Infrastructure Is the New Frontier

Edge AI is a way for a business to run “smart” software directly where work happens—on a device, a machine, or a local computer—rather than sending everything to a distant cloud first. In plain terms, it helps you react faster, keep more data on-site, and keep operations moving even when connectivity is patchy. A few simple examples make the idea more concrete: A small shop uses a local camera system to detect when checkout queues grow too long and alerts staff before customers start leaving. A factory adds a vibration sensor and a lightweight anomaly model to one machine, so unusual patterns are flagged before a breakdown causes downtime. A food distributor monitors cold storage locally and sends alerts only when temperature drift matters, instead of depending on constant cloud sync. These are not massive rebuilds. They are focused on operational improvements. In one minute Start with one operational bottleneck (queues, spoilage, missed faults, slow inspections). Pick a “local decision” that benefits from speed (approve/reject, flag/ignore, stop/continue). Pilot on a single site with a measurable target (less downtime, fewer stockouts, faster service). Keep humans in charge: Edge AI should recommend or flag before it automates. The problem → what changes... Edge AI in Everyday Operations

Introduction Honestly, this week felt different. Not because of another big model launch, but because the surrounding stories became harder to ignore. AI is no longer just changing what tools can do. It is changing how companies justify layoffs, how workers experience their jobs, and how model providers position themselves in the stack. GPT-5.4 matters. But the bigger signal this week is that AI is reshaping institutions, incentives, and trust at the same speed it reshapes software. These are not abstract signals. They affect how products get built, where value accumulates, and what work feels like for the people expected to supervise these systems. Of the eight signals below, three matter most: agentic tooling is consolidating, AI is changing workforce narratives faster than work itself, and trust is becoming a real market variable. Developer Tools and Models 1. GPT-5.4 launched on 5 March — and it changes how agents are built Introducing GPT-5.4 — OpenAI OpenAI launches GPT-5.4 with Pro and Thinking versions — TechCrunch If you have built agents recently, you have probably felt the friction of routing between a reasoning model and a coding model. GPT-5.4 addresses that directly. OpenAI merged GPT-5.2’s general reasoning and GPT-5.3-Codex’s coding depth... Better Models, Burnout, and a $599 Mac

Introduction My email inbox used to be a peaceful place. Then one day it became… a cosmic singularity of newsletters, receipts, notifications, and mysterious marketing emails from websites I swear I visited exactly once in 2014. You know the situation: 100,616 emails in total 47,234 unread emails “Important” emails hiding somewhere inside 47 newsletters about productivity that you never had time to read At that point, the inbox stops being a tool and starts becoming a guilt generator. The good news: cleaning it up is very possible. And you do not need to spend an entire weekend deleting emails like a medieval scribe sorting parchment. Here are practical, slightly nerdy, and sometimes AI-powered ways to take your inbox back. Step 1 — Accept the Truth: Most Emails Are Not Important The first psychological breakthrough: 80–95% of email is noise. Examples: Newsletters you skim once every six months “Special offer just for you!!!” Automated notifications Social media updates “We updated our privacy policy” emails Your inbox is not a museum archive. It’s a temporary processing system. So rule number one: If you wouldn’t search for it later, it probably doesn’t deserve to stay. Step 2 — The Fastest Manual Cleanup Trick... My Inbox Is a Black Hole (and How I Escaped It)

Introduction This week, the most important AI news was structural, not theatrical. Yes, there were launches—several significant ones. But if you step back, three forces are now moving in the same direction at the same time: model economics are compressing fast, inference infrastructure is being rebuilt from the ground up, and policy constraints are shifting from aspirational frameworks to operational reality. That combination changes the competitive landscape in ways a single model release simply cannot. The practical consequence: winning in AI is no longer about having the cleverest model. It is increasingly about deploying the right tier at the right cost, on infrastructure you actually control, within governance boundaries that are tightening whether you are ready for them or not. Major Product and Model Launches 1. Google launched Gemini 3.1 Flash-Lite for high-volume production workloads Google releases Gemini 3.1 Flash Lite at 1/8th the cost of Pro Gemini 3.1 Flash-Lite: Built for intelligence at scale Reuters-syndicated report on Gemini 3.1 Flash-Lite pricing and rollout On 3 March 2026, Google released Gemini 3.1 Flash-Lite—the latest in its Gemini 3 family, positioned as the fastest and most cost-efficient option in that line. VentureBeat frames the pricing relative to Gemini 3.1 Pro at... AI Is Splitting Into Tiers

Introduction Vibe coding is fun. You open an AI tool, describe an idea, and minutes later, you have working code. I built apps that way, too. Some worked. Most didn’t last. They were exciting experiments — but not reliable tools. Over time, I realised something uncomfortable: Vibe coding wasn’t enough. If I wanted apps that I actually used — apps that saved time, automated workflows, and ran reliably — I needed structure. A little vibe story I built an AI-powered tool in one evening. It felt magical — until it broke when I needed it most. I couldn’t explain the architecture, trace the changes, or roll back safely. It worked, but it wasn’t built to last. I rebuilt it with a clear problem definition, a spec, milestones, and Git discipline. The second version didn’t just run — it held up. That’s when I realised vibe coding wasn’t enough. AI can generate code in seconds — but without a structured AI coding workflow, it rarely produces reliable software. Two ways to build with AI Chaotic Vibe Coding Idea Prompt Code Patch Scope creep Abandon ↳ ends up in /git, untouched /git/cool-idea-v3/ /git/newsletter-app2/ /git/scraper-final-v9/ lost momentum. lost purpose. Structured AI Development Problem... Vibe Coding Wasn't Enough — The Lightweight System I Use to Turn AI Prompts into Deployed Apps

Introduction This was not a quiet week for AI — it just looked quiet on the surface. Underneath, something more interesting was happening: the technology and the world it runs on are starting to pull in opposite directions. On the software side, the pace is striking. Anthropic acquired Vercept to push Claude’s ability to see and operate software interfaces past the 72.5% mark on OSWorld — up from 15% just fifteen months ago. Cloudflare reimplemented 94% of a major web framework in a single week using Claude, for roughly the price of a cheap flight. Google launched Nano Banana 2 (Gemini 3.1 Flash Image) and Perplexity had it running inside their new multi-agent Computer tool on the same day — a day-zero integration that would have been unimaginable two years ago. The software layer is moving fast and integrating even faster. On the physical side, the signals tell a different story. Eight hyperscalers are on track to spend $710 billion on AI infrastructure in 2026 — and that capital race is already raising the price of RAM in consumer laptops and potentially shrinking the memory in your next budget smartphone. Power grids cannot keep up with the demand; hyperscalers are... 72.5%, $710B, and a March in London

Live Design Testing with GitHub Pages and a Custom Domain In my previous post, I wrote about using Git branches to test blog designs before committing to them. A few of you asked a natural follow-up question: what if you want to run two experiments at the same time, each at its own live URL, without touching your main blog at all? I had the same question! And it turns out GitHub Pages has a lovely little feature that makes this possible. The trick is not branches — it is repositories. How GitHub Pages Handles Custom Domains When you link a custom domain to your GitHub Pages setup, it applies to your entire GitHub account, not just one repository. GitHub distinguishes between two types of Pages sites: Your user site lives in a repository named exactly yourusername.github.io. This is where your main blog lives, and it publishes to the root of your custom domain — domain.com. Any other repository in your account becomes a project site, and GitHub Pages automatically publishes it as a subdirectory of your custom domain. So a repository called test1 publishes to domain.com/test1, and a repository called test2 publishes to domain.com/test2. No extra configuration needed —... Live Design Testing with GitHub Pages and a Custom Domain

Testing Blog Designs with Git Branches and GitHub Pages I have a habit of tweaking my blog design and then immediately second-guessing myself. Does this colour scheme actually work? Is this layout easier to read, or just different? For a long time, the only way I could really tell was to push the change live and look at it in a real browser — which felt a bit reckless. Then I discovered that GitHub Pages lets you publish from any branch, not just main, and everything changed. Here is the technique I now use whenever I want to compare two designs side by side before committing to one. Why Branches Are Perfect for This Git branches are usually associated with software features or bug fixes, but they are genuinely useful for design experiments too. The idea is simple: your main branch holds the version of your blog that is live and working. You create a new branch for your design experiment, push it to GitHub, and then tell GitHub Pages to serve from that branch instead. You get a real, live preview of your experiment — fonts, layouts, colours, everything — without touching your production site. When you are happy... Testing Blog Designs with Git Branches and GitHub Pages

How to Host Your Blog for Free with GitHub Pages A little while ago, a friend asked me how I host this blog. When I told her it was free, ran on GitHub, and I had never paid a hosting bill for it, she did not quite believe me. I completely understand that reaction — it sounds too good to be true. But GitHub Pages is genuinely one of the best-kept secrets for anyone who wants a personal blog or website without the overhead of managing a server or paying for hosting every month. In this post I want to walk you through everything from scratch: creating a GitHub account, setting up your blog with Jekyll, choosing a theme, writing your first post, and connecting a custom domain if you have one. Whether you have never used Git before or you are a developer who just has not gotten around to setting up a personal site, I hope this gives you everything you need to get started. What Is GitHub Pages, Exactly? GitHub Pages is a free hosting service built into GitHub. You store your website files in a GitHub repository, and GitHub automatically builds and serves your site at... How to Host Your Blog for Free with GitHub Pages

Before you install it locally, here are five entirely plausible ways your week could take an unexpected turn. It looks harmless at first. You connect OpenClaw to your Gmail. You point it at Slack. You give it a few instructions and step away to make coffee. But the moment it can read your inbox, post on your behalf, and call external APIs with your credentials — something changes. The moment a system can act on your behalf with real credentials and persistent consequences, it becomes infrastructure. And infrastructure, as I have learned, has very different rules. This post complements this week’s AI Signals, where I examine the broader capability, capital, and sovereign investment shifts shaping agentic AI at scale. Introduction AI agents like OpenClaw are wonderful — genuinely exciting tools that we are only beginning to understand. Unlike a chatbot, OpenClaw can monitor Slack channels, read and draft Gmail messages, call external APIs, execute structured workflows, and trigger automated actions. It is not answering questions. It is acting on your behalf, in your name, with your access. That distinction matters enormously — and most people miss it entirely until something goes wrong. Five Ways a Local Install Can Ruin Your... OpenClaw Isn't a Chatbot Anymore. It's Infrastructure.

Git is usually about managing streams of history. You merge rivers of code together. But sometimes, you don’t want the whole river. You just want one specific fish. That’s git cherry-pick. Introduction The command git cherry-pick <commit-hash> takes the changes from a single specific commit and applies them to your current branch as a new commit. It’s highly useful in specific scenarios: Hotfixes: You fixed a bug in dev, and you need that same fix in prod right now, but you can’t merge dev yet because it has unfinished features. Mistakes: You committed to the wrong branch. How to use it Let’s say you are on main and you want a commit abc1234 that exists on feature-branch. Find the hash: git log feature-branch # verify that abc1234 is the one you want Execute the pick: git checkout main git cherry-pick abc1234 Git will take the diff from that commit and try to apply it to main. If successful, it automatically creates a commit. Handling Conflicts Just like a merge, cherry-picking can result in conflicts if the code has diverged too much. If this happens, Git will pause. Open the conflicted files and resolve the <<<< markers. git add <file> git... Git Cherry-Pick: The Surgeon's Knife

Introduction What a week this has been! Between February 12 and 19, 2026, three very different layers of the AI world moved at the same time: major model releases landed (Claude Sonnet 4.6 and Gemini 3.1 Pro), a staggering amount of capital was raised ($30B Series G), and a national research body published a funded strategy (UKRI’s £1.6 billion plan). I found the combination fascinating, so let me walk you through what happened, why it matters, and what I think it means for developers. 1. Anthropic Released Claude Sonnet 4.6 (Feb 17, 2026) Anthropic: Introducing Claude Sonnet 4.6 On February 17, Anthropic released Claude Sonnet 4.6, and it is not a minor update. The headline improvements are stronger coding support, better computer-use capabilities, and more reliable agent planning — all backed by a 1 million token context window. To put that in perspective, 1 million tokens is roughly 750,000 words, which means Sonnet 4.6 can reason across entire codebases or long document collections in a single pass without losing earlier context. Market Reaction & Independent Coverage Anthropic releases Sonnet 4.6 TechCrunch covered the release on the same day and made a point I agree with: this is not a quiet... Agentic AI at Scale: New models, $30B, and the UKRI Strategy

I am lazy. But usefully lazy. I believe that if you type the same long command more than three times a day, you should shorten it. Git commands can be verbose, but Git has a built-in aliasing system to fix that. Introduction Aliases allow you to map short commands to longer Git functions. You define them in your .gitconfig file (usually in your home directory) or by using the command line. Here git config --global alias.co checkout allows you to type git co instead of git checkout. Here are my top essential aliases. The Essentials Run these in your terminal to set them up: # Basic navigation git config --global alias.co checkout git config --global alias.br branch git config --global alias.ci commit git config --global alias.st status # Unstaging files (undoing 'git add') git config --global alias.unstage 'reset HEAD --' Now, checking status is just git st. It saves milliseconds, but they add up to a feeling of fluidity. The Visualizer: A Better Log The default git log is a bit dry. I use an alias called lg that makes it look like a rainbow graph: git config --global alias.lg "log --color --graph --pretty=format:'%Cred%h%Creset -%C(yellow)%d%Creset %s %Cgreen(%cr) %C(bold blue)<%an>%Creset' --abbrev-commit"... Type Less, Do More: My Top 10 Git Aliases

Introduction This week felt like watching two forces pull against each other. On February 7th, both OpenAI and Anthropic released advanced models simultaneously. ByteDance launched Seedance 2.0 with video quality that made Elon Musk say it is happening “too fast.” Modal Labs is raising at a $2.5B valuation. Perplexity is running three frontier models in parallel to cross-validate answers. The capability momentum is real. But the friction from the real world is getting louder. Data centre projects are stalling in permit review. Communities are organising opposition. Microsoft is betting on speculative superconductor technology because conventional power delivery cannot scale. OpenAI changed its mission alignment team. ByteDance’s Seedance 2.0 launched with certain real-person content generation features limited or paused due to privacy and misuse concerns. What stood out to me most is that the conversation is shifting. It is not only about model capability anymore. It is increasingly about who gets power, who bears costs, and who keeps control. These are harder questions, and they do not have clean technical solutions. 1. Inference Infrastructure Funding Momentum Continues AI inference startup Modal Labs in talks to raise at $2.5B valuation, sources say TechCrunch reports that Modal Labs is in talks for a... AI Improves Itself While We Argue About Permits

Humans are terrible at repetitive tasks. We forget to run linters. We forget to check for trailing whitespace. We forget to run the test suite before pushing. Computers, on the other hand, love repetition. This is where Git Hooks come in. Introduction Git Hooks are scripts that Git executes before or after events such as: commit, push, and receive. They allow you to “hook” into the Git workflow and stop bad things from happening. The most useful one for daily development is the pre-commit hook. What is a pre-commit hook? This script runs every time you type git commit. If the script exits with an error (non-zero status), the commit is aborted. It’s the perfect place to put comprehensive checks: “Does the code compile?” “Are there any lingering print() statements?” “Does the formatting match the team style?” Setting it up (The Easy Way) You can write bash scripts in .git/hooks/, but that’s hard to share with a team. I recommend using the pre-commit framework. Install it: pip install pre-commit Add a config file .pre-commit-config.yaml to your repo: repos: - repo: https://github.com/pre-commit/pre-commit-hooks rev: v4.4.0 hooks: - id: trailing-whitespace - id: end-of-file-fixer - id: check-yaml Install the git hook scripts: pre-commit install... Git Hooks: The Robot Butler for Your Code

This is Part 2 of the Codex CLI series. Today, we’ll learn how to control Codex’s capabilities and make your first safe edits. This post is about speed with control, not automation for its own sake. Introduction In the last post, we installed Codex CLI and took our first steps with read-only exploration. Today, we go deeper: learning how to control what Codex can do, understanding the essential commands you will use daily, and making your first actual code edits—safely. What I learned from using Codex is that the key to productive work is understanding the control mechanisms. Unlike a chatbot that only gives suggestions, Codex can actually change your files and run commands. That power is valuable, but it requires proper guardrails. Let me show you how to stay in control while getting real work done. Understanding Security Controls: Permissions and Approvals This is the most important concept to understand before you let Codex make any changes. Codex uses a security system built on two interconnected ideas: permissions and approvals. At first, these might seem like the same thing, but understanding the distinction helps you think clearly about how Codex operates and how to stay in control. Permissions vs... Codex CLI Part 2 — Security Controls & Safe Editing

Introduction This week I observed something curious. AI is advancing faster than ever, yet the physical world continues to set the pace. It reminded me of watching two runners on different tracks — one sprinting effortlessly, the other climbing uphill with a heavy backpack. Many of this week’s signals point to the same tension: software speed versus physical limits. Here are the stories that made that contrast feel especially sharp. 1. AI-Assisted Cloud Break-Ins Are Now Measured in Minutes Intruder uses AI assistant in AWS cloud break-in A Sysdig security report described an attacker achieving administrative privileges in under ten minutes, moving from stolen credentials to AWS Lambda execution. LLM-generated code was used to accelerate the process, and investigators noted artefacts consistent with machine-assisted scripting rather than purely human-written tooling. Why This Matters AI is collapsing the time between access and impact. Security assumptions built around slow, manual attackers no longer hold. Detection alone is insufficient when adversaries can chain complex steps together in minutes with machine assistance. Response speed now matters as much as prevention. 2. Power Queues in Europe Are Now Multi-Year Bottlenecks Amazon says European data center power can take seven years to connect AWS executives warned... The AI Paradox: Lightning Fast and Gridlocked

While polishing my publishing script, I managed to do the one thing I explicitly advise against: I committed a token to Git. It was in a comment. It was in a private repository. It was still, regrettably, committed. What followed was not drama, but administration — rewriting history, checking remote branches, and searching old commits for fragments of the token to ensure it had truly vanished. It turns out that removing a secret from Git is rather more involved than removing a semicolon. On balance, I would not recommend the experience. It was, however, a useful reminder that secure workflows are not theoretical best practices. They are habits — and habits are most valuable when we are tired, moving quickly, or feeling slightly too confident. Introduction There is something very addictive about modern code assistants, and I find myself using them almost daily. The efficiency gains and faster prototyping are obvious on the surface. What continues to amaze me is how well AI assistance understands what we want to implement, often from very small or loosely defined specifications. You type a half-formed thought — “parse this CSV”, “add authentication”, “why does this crash?” — and suddenly there is structure, clarity,... Using AI Code Assistants Safely

This week’s AI news was quietly consequential, and I found myself thinking about what these developments mean for the field I care so much about. Instead of flashy new demonstrations or larger models, the important signals appeared in earnings calls, export rules, shipping approvals, and security reports. Microsoft tied AI directly to long-term capital spending. Anthropic argued for regulation centred on chip access. China approved limited H200 imports. And at the other end of the technology stack, desktop compute and open models continued to advance — alongside significant security friction that caught my attention. None of these stories is flashy on its own. But together, they paint a picture of AI settling into infrastructure: budgeted, gated, and increasingly operational. Let me share what stood out to me this week. 1. Microsoft Earnings Put AI Capex Front and Centre Microsoft investors sweat cloud giant's OpenAI exposure Microsoft reported $81.3 billion in revenue for Q2 FY2026, a 17% year-over-year increase and higher than analysts’ expectations — with Microsoft Cloud revenue alone surpassing $50 billion. These results are directly linked to continued demand for artificial intelligence services and to investment in cloud infrastructure. (See Microsoft beats Wall Street expectations with $81.3B revenue.) Despite... Chips, Capex, and Code Risk

We’ve all seen it. You clone a repository, and there it is: a .DS_Store file, a __pycache__ folder, or worse—a file containing local API keys. Committing these files is messy, unprofessional, and sometimes dangerous. The solution is simple but often misunderstood: the .gitignore file. Introduction The .gitignore file tells Git which files it should intentionally ignore. It’s not just about keeping your repo tidy; it’s about security and collaboration. You don’t want to force your local editor settings or operating system junk on your teammates. The Basics Create a file named .gitignore in your project root. Add patterns for files you want to exclude: # Dependencies node_modules/ .venv/ # Compiled code *.pyc __pycache__/ dist/ # System files .DS_Store Thumbs.db Now, git status won’t even show these files. Global Ignore: The files you ALWAYS ignore Some files, like macOS .DS_Store or editor configurations (.vscode/), haunt every project you touch. Instead of adding them to every single project’s ignore file, you can set up a Global .gitignore. Create a file at ~/.gitignore_global. Add your OS-specific junk there. Tell Git to use it: git config --global core.excludesfile ~/.gitignore_global Now, you never have to worry about accidentally committing a .DS_Store file again. Fixing Mistakes:... Stop Committing Garbage: A Masterclass in .gitignore

TL;DR: This week felt less like model drama and more about the systems shaping what AI can actually do. Courts, export rules, and commerce protocols are becoming as important as the models themselves. Here are five signals that stood out. Introduction You know what? Sometimes the most interesting AI developments have nothing to do with new models or benchmark scores. This week reminded me of that. Whilst everyone obsesses over the latest transformer architecture or chatbot capabilities, the real story is happening in courtrooms, congressional committees, and standards bodies. I find this fascinating because it mirrors something I’ve observed throughout my career in computer science: the technical capabilities matter far less than the systems and rules that govern how we can use them. It’s like learning to code—you can master Python syntax, but if you don’t understand the broader ecosystem, licensing, and community standards, you’re missing the bigger picture. So let’s dive into this week’s signals. Fair warning: there’s a minor timing issue I need to address upfront about one of these stories, but I’ll explain that when we get there. Five Signals That Actually Matter Weekly AI Signals: Key Takeaways Signal Industry Impact Builder Action PwC CEO Survey: AI... This Week in AI: Regulation Heat, Cloud Bets, and Agentic Shopping

Have you ever been in the middle of a complex feature, writing messy code, when suddenly a colleague asks you to fix a critical bug on production? You can’t commit your broken code. You can’t lose your work. So, what do you do? You stash it. Introduction git stash is one of those commands that feels like magic once you start using it. It takes your uncommitted changes (both staged and unstaged), saves them away for later use, and reverts your working directory to the last clean commit. Think of it as a “Cut and Paste” for your entire project. You cut your current work, go do something else, and paste it back when you’re ready. Git Stash Command Reference Command What It Does git stash Stash all tracked changes (staged + unstaged). git stash -u Stash tracked and untracked (new) files. git stash pop Apply latest stash and remove it from the list. git stash apply Apply latest stash but keep it in the list. git stash list View all stashed entries with timestamps. git stash pop stash@{1} Apply a specific entry by index. git stash drop stash@{0} Delete a specific stash entry. git stash clear Delete all stash... Git Stash: The CTRL-Z for Your Working Directory

This is Part 1 of the Codex CLI series. We’ll cover installation, authentication, and your first session. Future posts will explore workflows, best practices, and advanced features. Introduction Today I want to introduce you to Codex CLI—a tool that has genuinely changed how I work with both code and writing. If you have ever wished you could have an AI assistant that actually understands your project, can read your files, and help you make changes right from your terminal, Codex CLI is exactly that. It is not just another chatbot; it is designed to work inside your repository, understanding your code structure and helping you improve it safely. In this series, I will walk you through everything you need to know to use Codex CLI productively. Today’s post covers the fundamentals: what it is, how to install it, and how to take your first steps safely. What Is Codex CLI? Before we jump into installation, let me explain what Codex CLI actually is and why it is different from using ChatGPT in a browser. The Core Concept Codex CLI is a coding agent that runs locally in your terminal. When you start it in a project directory, it can: Read... Getting Started with Codex CLI

This week felt grounded. Instead of big model claims, the signals were about governance, power, and the plumbing that lets AI show up in the real world. The quiet shift is toward infrastructure you can point to and systems you can live with. Here are the five signals that made that shift feel concrete. Weekly AI Signals: Key Takeaways Signal Industry Impact Builder Action State AI Safety Laws California and New York are advancing binding AI safety frameworks, creating de facto national standards. Integrate audit trails and documented model governance into your products now; compliance will become a baseline expectation. Open Model Adoption Gap Despite proven technical benefits, open models still face significant enterprise adoption friction. Invest in LLM deployment tooling (inference servers, fine-tuning pipelines) to lower barriers and avoid vendor lock-in. Texas as AI Infrastructure Hub Power availability and land access are making Texas a strategic data-centre hotspot, reshaping regional pricing. Factor geographic diversity and energy sourcing into infrastructure decisions; single-region deployments carry growing risk. Meta GenAI Data Flow Design Competitive edge is shifting from model quality alone to data pipeline efficiency and auditability. Model data flow orchestration as a first-class engineering concern, not a secondary operational task. Micron... The Week AI Got Practical: Laws, Power, and Open Models

Introduction This is a very reasonable question — and you’re not alone in feeling puzzled. Right now we have many powerful tools, but each one shines in a different layer of work. The confusion usually comes from trying to make one tool do everything. Let’s untangle this calmly and then build a clear, future-proof mental model for you. I’ll do this in four parts: What Codex CLI is really good at (and what it isn’t) How it compares to MCP, Antigravity, and Cursor Which tool fits your blogging + Python workflow best How to design a multi-agent workflow without chaos I’ll keep this practical, not hype-driven. 1. Will Codex CLI fit you well? Short answer Yes — but as a worker, not as the brain. Codex CLI is best thought of as: A careful, local execution agent that lives next to your files It is not: a long-term planner a creative ideation engine a memory system It is: excellent at editing Markdown safely great at refactoring Python good at running tests, linters, and checks strong at repeatable, terminal-based tasks For blogging, Codex CLI fits the “make this concrete” phase. 2. How Codex CLI compares to MCP, Antigravity, and Cursor Here’s... My Multi-Agent Workflow

This week in AI felt a little different to me. Fewer headlines about dazzling benchmarks or clever prompts — and more about where AI actually lives, who powers it, and how it starts to touch everyday systems. What I found interesting is that all four stories below point in the same direction — not toward new capabilities, but toward where AI is settling in the real world. Chips, electricity, assistants we already talk to, and even shopping flows. Less magic. More plumbing. And that’s often where the most important shifts begin. Weekly AI Signals: Key Takeaways Signal Industry Impact Builder Action TSMC $52–56B Capex AI accelerator demand is structural, not cyclical; TSMC forecasts mid-to-high-50% CAGR in that segment through 2029. Stop treating AI compute as elastic cloud burst capacity; plan for hardware availability windows and supply chain lead times. Apple + Google Gemini Apple chose to integrate Gemini models as the foundation for Apple Intelligence and Siri rather than build in-house. Large-scale model partnerships will define platform AI capabilities; build integration layers that can swap model backends. Microsoft Community Infrastructure Data-centre electricity demand will more than triple by 2035; infrastructure is now an urban planning and grid concern. Factor data... Signals from the AI Supply Chain

Weekly AI Signals: Sovereignty, Safety, and Physical AI Limits (Week of January 9, 2026) Hello, Dear Reader! This week in AI felt noticeably different from recent months—quieter, but in a way that felt more meaningful rather than less important. Instead of louder models or bigger capability announcements, the conversations shifted to constraints: where AI actually runs, who controls it, what happens when deployment races ahead of safety, and how AI performs when mistakes are genuinely unacceptable. Less spectacle, more reality. What I found interesting is how these stories connect. They are all, in different ways, about limits—technological, geopolitical, ethical, and physical. After months of “what can we build?” we are seeing more questions about “under what conditions should we build it?” Here are six developments from this week that I think reveal where AI is heading next. Weekly AI Signals: Key Takeaways Signal Industry Impact Builder Action France + Mistral Sovereignty Governments treat AI as critical national infrastructure requiring jurisdictional control, not a cloud commodity. For public-sector clients, design for data residency and local deployment from day one; technical capability alone is not sufficient. Nvidia: Training → Inference Jensen Huang confirms the industry bottleneck has shifted from training to deployment,... AI's Week of Limits: Safety, Control, and Real-World Physics

This post is part of my Weekly AI Signals series—a curated look at the moments that matter once the noise fades. Weekly AI Signals: Regulation, Infrastructure, and Autonomy in Late December 2025 As we step from 2025 into 2026, I want to pause on the final week of the year — not because it was loud, but because it was revealing. Three developments stood out. China released draft rules aimed at emotionally engaging AI systems. SoftBank moved to strengthen its position in digital infrastructure. And Meta acquired a company focused on autonomous AI agents. Individually, these stories are fascinating. Taken together, they suggest something deeper: AI is moving into a phase shaped by governance, physical scale, and questions of agency. Let’s walk through what happened — and what it might mean for where we’re headed. Weekly AI Signals: Key Takeaways Signal Date Industry Impact Builder Action China Emotional AI Rules 27 Dec 2025 Regulators treat relational AI UX as a distinct safety risk, requiring transparency, break prompts, and human review for self-harm signals. Design AI companions with visible disclosure labels and usage-time management as a first-class feature, not an afterthought. SoftBank acquires DigitalBridge 29 Dec 2025 $4B deal confirms that... As 2025 Closes: AI's Week of Regulation, Infrastructure, and Autonomy

This post is part of my Weekly AI Signals series — a curated look at the moments that matter once the noise fades. Weekly AI Signals: Five AI Developments From Late December 2025 Five Signals That Mattered Hello, dear reader! Welcome to the last week of December 2025. I hope you are enjoying the holidays and have had a moment to look back on what has been an extraordinary year for AI. This is not a complete account of everything that happened in AI this week. Instead, it is a small, curated set of signals that felt meaningful once the noise settled — moments where limits became visible, incentives shifted, or assumptions quietly changed. If 2025 was the year we kept asking “what can we build?”, this past week felt like the moment the industry started asking a more useful question: “what actually works?” The five signals below come from very different places — creative industries, hardware, developer practice, security, and physical systems — but together they point to the same thing. AI is moving out of its novelty phase and into an engineering one. Here is what stood out, and why it may matter longer than this week’s headlines.... Hardware Handshakes, Prompt Injection Reality, and AI Beyond the Screen

Merry Christmas & a Happy New Year 🎄 I wish you joyful moments with your loved ones. Have a prosperous and happy 2026! I genuinely appreciate your visit to my blog, and I’m thrilled when I hear it’s been helpful to you. Many of you are skilled coders and experts in your fields, and I wish you great success—not only in 2026 but also in the many happy years ahead. Doing something well energises our lives in a way no AI can replicate. I hope you feel inspired about your work this year and enjoy exploring new techniques and AI tools. AI is a powerful tool that can enrich our lives and make us more productive, ultimately saving what matters most: our time. I’ve found that AI can save tremendous time when you know exactly which tools to use and how to use them effectively. That’s why I’ve shared my favourite AI tools on the blog, along with practical guides for using them. You can check it out here. While generative AI can answer technical questions and write code in various styles, it’s important not to lose the human touch. Generative AI represents a productivity tool that accelerates technical and... Merry Christmas and a Very Happy New Year!

This Week’s AI Signals: Multimodal Models, Safety Defaults, and Agent Interfaces This week, several AI developments caught my attention. Not because they were particularly loud or novel, but because they touched on questions that tend to surface later, when systems are already in use. Better safety defaults are one of those questions. If AI systems are going to be used by children and teenagers, safety cannot remain an afterthought or a policy document. It needs to be part of how applications are designed from the start — even if that means slower progress or fewer features. Alongside this, we saw continued movement toward faster, agent-ready models and interface tooling that treats interaction as something adaptive rather than static. None of these developments are dramatic on their own. But together, they hint at where current AI systems are under pressure to change as they move closer to everyday use. Weekly AI Signals: Key Takeaways Signal Industry Impact Builder Action Meta Mango (Multimodal) Unified image/video/text reasoning models replace fragmented pipelines, simplifying architecture. Design applications with multimodal I/O as a first-class concern rather than an optional add-on. OpenAI & Anthropic Safety Child and teen safety transitions from a policy document to a structural... AI Interfaces, Safety, and Multimodal Systems

Antigravity 1.11.9 vs Cursor 2.1.42 (Universal) Two IDEs. Two philosophies of AI-assisted coding. Google’s Antigravity and Cursor are both AI-powered IDEs, but the way they help a developer think and work is very different. In this piece, I compare them head-to-head and link to official documentation or changelogs so you can explore the exact features I describe. Google Antigravity 1.11.9 Outcome-oriented, agentic development IDE Official site: https://antigravity.google/ Developer guide: Build with Google Antigravity (developers.googleblog.com) Getting started tutorial: https://codelabs.developers.google.com/getting-started-google-antigravity Antigravity is an agent-first development platform that delegates entire task workflows — planning, coding, testing, and verification — to autonomous agents operating across the editor, terminal, and browser, rather than completing code line by line. Antigravity’s Agent-Oriented Workflow Imagine a development environment that says: “Tell me your goal. I’ll handle the workflow.” Agents can run across your editor, terminal, and browser — not just suggest text in a sidebar. Antigravity Feature Set: Agent Manager, Artifacts, Multi-Model Support Agent Manager & Mission Control — A dashboard to run and monitor multiple AI agents handling parts of a project in parallel. Artifacts — Agents produce verifiable outputs like task lists, implementation plans, screenshots, code diffs, and browser recordings so you can see what changed... Antigravity 1.11.9 vs Cursor 2.1.42 (Universal): A Practical Comparison

Weekly AI Signals: Robotic Labs, Ambient Hardware, and AG Safety Demands This week, AI edged a little further into the physical and infrastructural world. DeepMind is setting up its first automated materials science lab in the UK. OpenAI has completed early prototypes of its new ambient hardware device — something deliberately quieter and more context-aware than today’s screens. And in the US, 42 attorneys general have made it clear: unsafe chatbot behaviour is no longer something companies can simply promise to improve “later”. Alongside these stories, a major $20 billion AI infrastructure partnership was announced, and new findings showed where AI tools already rival human specialists. Here is what mattered this week — and why it shapes the systems we build. Weekly AI Signals: Key Takeaways Signal Industry Impact Builder Action DeepMind Robotic Labs AI moves from digital tokens to physical manipulation, accelerating materials science discovery. Abstract planning architectures (LLM to physical action) will soon be standard in manufacturing; learn ROS alongside Python. OpenAI Ambient Hardware Hardware pivots from glowing rectangles to screenless, context-aware auditory/environmental sensors. Prepare for UI-less software engineering where voice and context state replace traditional DOM rendering. State AGs Demand Safeguards 42 U.S. Attorneys General signal that... Labs, Law and New Hardware Horizons

I went to the dentist today. You know, the adult version of a school exam, except the chair is oversized and the lighting is uncomfortably good. The hygienist always asks “Any concerns?” and suddenly my brain goes blank. Which teeth do I even have? Where are they located? What is a molar? They tilted the chair back, switched on that tiny headlamp of truth, and my soul decided to take a brief walk around the waiting room. Many years ago I lost a tooth to a small stone hiding in my food. Today I am finally getting a new one — not real, but perfect. 🦷 Please wish me luck. :) Also, considering all the dental suffering in human history, where are the AI dentists? Surely robots could make this process less terrifying. Or at least tell better jokes while drilling. AI and Robotics in Dentistry: What Changes in the Next 2–7 Years After today’s adventure, I got curious: what is actually happening in dental tech while we are all lying back practicing controlled breathing? AI-assisted dentistry is a category of clinical practice that uses machine learning models to support diagnosis, risk prediction, and robotic-assisted procedures alongside — rather than... A Short Tale of Bravery (at the Dentist)

Weekly AI Signals: AGI Timelines, 3D Vision, and Voice-Cloning Scams This week, the theme was convergence—but with a side of caution. We saw the convergence of policy and technology as the U.S. Health Department moved to make AI part of its core infrastructure. We saw the convergence of senses, with breakthroughs in how AI sees (3D from 2D) and hears (universal sound understanding). But we also saw the convergence of AI capabilities and criminal intent. While DeepMind predicts AGI by 2030 and researchers give machines better senses, a story out of Kansas served as a chilling reminder of why we need to stay vigilant right now. Weekly AI Signals: Key Takeaways Signal Industry Impact Builder Action U.S. HHS Strategy Government transitions from experimental pilots to operationalising AI as core infrastructure. Prepare software for rigorous federal compliance and data privacy standards if selling into the public sector. AGI by 2030 DeepMind shifts focus from LLMs to “world models” to achieve AGI within five years. Begin architecting systems that can interface with predictive, physics-aware models, not just text generators. The “Trust Moat” Western AI companies leverage commercial trust as a geopolitical advantage over foreign competitors. Double down on transparent data governance to... AGI Timelines, 3D Vision, and the Reality of AI Scams

Hello, my Dear Reader, We are celebrating this blog’s birthday again! Elena’s AI Blog is now four years old—still learning, still growing, and still navigating this fascinating AI landscape together with you. It has been an incredible year since I wrote Three years of Elena’s AI Blog. The AI world has moved at breathtaking speed, and I have been here, learning alongside you, documenting the journey, and sharing my thoughts on everything from multimodal AI to coding assistants. What is Elena’s AI Blog? Elena’s AI Blog is a personal technology blog that documents hands-on testing of AI coding assistants, multimodal models, and machine learning fundamentals for a general technical readership. Like everyone today, I live in an era of rapid AI evolution. It is challenging to understand and live in, even for people with a technical background. However, I love making things easy to understand while learning new technologies as a passion. I created this blog to log what I learn and share my ideas and findings. Now four years old, this blog continues to connect technology with everyday understanding, reflecting my passion for coding and commitment to making complex concepts accessible. The Blog Since December 2024 Since my last... A Journey Through AI and Code

The New Developer Skill Stack: From Writing Code to Managing AI Agents Last week, I built an app without writing a single line of code. It still feels slightly illegal to admit that out loud. The IDE stitched most of it together. The agents filled in the logic. I spent my time describing what I needed — like guiding a very enthusiastic intern who occasionally rewrites your entire project because it “felt cleaner.” And that’s when it hit me: developers aren’t disappearing — but what we do each day has already changed. I’m not laying bricks anymore. I’m the architect who guides the builders. Less typing, more thinking. Less wrangling syntax, more designing boundaries. Here’s what this new skill stack feels like in practice, with the real mistakes and odd surprises included. The New Hard Skills: Orchestration & Specification Intent Specification (Vibe Coding) Intent Specification, also called Vibe Coding, is the practice of writing precise natural-language contracts that constrain what an AI coding agent is allowed to build, preventing scope creep and unwanted files. Last Tuesday, I said, “Make the login more secure.” The agent returned something that looked like a spaceship airlock. Beautiful, impenetrable, and completely unusable. Agents are... The New Skill Stack, from Writing Code to Managing Intelligence

This Week’s AI Releases: Claude Opus 4.5, ChatGPT Shopping, EV Forecasting, and DeepSeekMath-V2 This week, AI didn’t make a fuss. Instead, it quietly slipped into places where it can genuinely help: in our editors, in our browsers, and even at the roadside charger. It’s the sort of progress that whispers, not shouts. Weekly AI Signals: Key Takeaways Signal Industry Impact Builder Action Claude Opus 4.5 Core reasoning capabilities upgraded with native Excel and Chrome integration, focusing on zero-hallucination execution. Use Opus 4.5 for complex multi-file codebase refactors where context tracking is more critical than pure speed. ChatGPT Shopping AI transitions from passive answering to active, agentic product research and curation. E-commerce developers must prepare structured semantic metadata, as users will increasingly search via LLMs rather than standard query bars. Google EV Forecasting Google proved that a simple linear regression model can outperform heuristics without the overhead of massive neural networks. Stop defaulting to LLMs for simple predictive tasks; lightweight deterministic models are cheaper, faster, and often better suited for edge deployment. DeepSeekMath-V2 An open-source model achieved gold-medal-level IMO scores via a rigorous generator-verifier loop. Implement self-verification loops in your own agentic workflows to dramatically increase output reliability on complex logical... Claude Opus, ChatGPT Shopping, EV Forecasting and DeepSeekMath-V2

Gemini 3, Antigravity, and Humanity-Centred AI: This Week’s Signals This week wasn’t just about new models. It was about growing up. Google and OpenAI delivered the expected fireworks: Gemini 3 refined the “Mixture-of-Experts” architecture for massive scale, and Project Antigravity killed the text editor in favour of agent orchestration. But the real signal didn’t come from a server farm. It came from the “adults in the room.” The WHO issued a strict mandate that “Humanity must hold the pen,” citing dangerous error rates in AI diagnosis. Ernst & Young demanded we start measuring the energy cost of intelligence (~1Wh per query). And on 60 Minutes, Anthropic’s CEO publicly questioned the unchecked power of unelected tech leaders—including himself. We are shifting from “look at this cool demo” to “how do we actually live with this?” The era of moving fast and breaking things is over. Welcome to the era of integration. If you're still waiting for AI winter, I have bad news: we're in AI summer, and nobody brought sunscreen. 1. Google Gemini 3 + “Antigravity”: The Death of the Text Editor? Google launched Gemini 3 on Tuesday. The model itself is impressive, but the real story is the environment it... Ethics, Gravity, and the Future We're Actually Building

Weekly AI Signals: Ethics, Hardware, and Developer Trust Some weeks in AI are loud and dramatic, while others offer a more subtle experience—a gentle reminder to notice interesting developments. This week was one of those softer moments, with seven noteworthy events that prompted me to reflect: We are truly building something new. This week showcases the connections between ethics, coding, hardware, and humanity. Weekly AI Signals: Key Takeaways Signal Industry Impact Builder Action Vatican AI Medical Ethics Major institutions are establishing formal ethical frameworks for AI deployments in healthcare. Ensure your medical AI pipelines have human-in-the-loop dignity checks, not just statistical accuracy tests. 9% Developer Trust in AI Blind trust in AI generation remains incredibly low; rigorous code review is non-negotiable. Enforce CI/CD pipeline checks for all agent-generated code to catch hallucinations before production. MIT LLM Modularity Structuring code around concepts and semantic boundaries to accommodate LLM reasoning. Write code as discrete, legible concepts rather than entangled logic blocks so agents can parse it effectively. NVIDIA DGX Spark A 1-petaflop workstation democratises massive compute, bringing datacenter power to the desk. Shift exploratory model training and inference testing back to local hardware to save cloud costs. VUNO Profitability Clinical AI transitions... Ethics, Code, Chips, and a Petaflop on Your Desk

AI and Religion: Historical Parallels to Institutional Resistance Every generation faces a moment when something new arrives—too big to ignore, too unfamiliar to immediately embrace. Centuries ago, that “new thing” was the printed book. Today, it may be artificial intelligence. In this blended reflection, I would like to explore two intriguing ideas: How the Church moved from resisting novelty to shaping AI ethics, and Whether AI itself could become a “new religion” for some. Let’s ponder it together. Historical Resistance to Disruptive Technology: Galileo, the Printing Press, and AI History gives us vivid examples of how disruptive new knowledge once felt. Historical Parallels: Technology vs. Institutions Historical Disruption Institutional Reaction The Core Fear Heliocentric Model (Galileo) Tried and found guilty of supporting ideas declared “formally heretical” (Galileo affair). Displacing humanity from the literal centre of the universe. The Printing Press The creation of the Index Librorum Prohibitorum (1559–1966) to ban “dangerous” texts. Democratised access to information undermining absolute authority. Astronomical Science (Copernicus) Forced to present his findings merely as hypothetical models rather than literal truth. Scientific observation contradicting established dogma. Imagine handing your code to a council of theologians and hearing: “This function feels suspicious.” But beneath the humour, there’s... Could AI Become a New Religion?

What Are Apache-Licensed Summarization Models? Apache-licensed summarization models are transformer-based NLP models distributed under the Apache 2.0 license, which permits commercial use, modification, and redistribution without royalty or disclosure obligations. You know what’s frustrating? Finding a brilliant AI model that summarises text beautifully, only to discover the license says “research purposes only” or worse — some vague terms that would make your lawyer cry. I spent way too much time digging through Hugging Face, reading license files, and testing models that claimed to summarize but just… didn’t. Most transformer models come with restrictive licenses that make you wonder if even looking at the model card might violate some terms. But here’s the good news: Apache 2.0-licensed summarization models exist. Real ones. Models you can actually use, modify, and ship in your apps without legal nightmares. I found them, tested them, and now I’m sharing them with you. Let’s dive in. Fun fact: I initially wanted to call this post "License-Free Summarizers" until my lawyer friend reminded me that "license-free" is a licensing nightmare in itself. Apache 2.0 it is! NLP Summarization Model Concepts: Transformers, BART, and T5 Before we jump into models and code, let’s quickly cover some terminology. Don’t... Apache-Licensed Summarizers

Hello, Dear Reader — how are you doing today? This week in AI, I wanted to focus on what actually matters for us developers. You know, the things that will make our lives easier (or at least more interesting) rather than just another hype cycle. So grab your favourite beverage, and let’s dive into five developments that might actually change how we work. Weekly AI Signals: Key Takeaways Signal Industry Impact Builder Action OpenAI’s $38B AWS Deal Signals a definitive shift to multi-cloud hyperscale, resulting in massive GPU availability by late 2026. Abstract your LLM API calls behind agnostic interfaces (like LiteLLM or LangChain) to swap providers easily as price wars heat up. Google Vertex AI Updates AI agent deployment moves from experimental sandboxes to production-grade, observable microservices. Implement formal agent telemetry tracking tokens, latency, and success criteria using Google’s Agent Engine. Copilot Org Governance IDE automation is no longer a wild west; enterprise policies can now dictate AI coding style at scale. Deploy a 10-line “house rules” document mapping your lint and testing standards directly into Copilot’s organisational settings. VS Code Unified Agents AI shifts from a sidebar widget to a first-class citizen deeply integrated into IDE planning and... AI Weekly — Agents Grow Up, Clouds Get Bigger

What Is Cursor 2.0? Cursor 2.0 is an AI-native code editor that replaces file-level autocomplete with autonomous, multi-file coding agents coordinated through a built-in model called Composer. Cursor 2.0 launched on October 29, 2025, and I am still figuring out whether Cursor 2 is right for my projects. If you’re wondering whether this upgrade is worth your time (and learning curve), here’s a clean, honest look. TL;DR: Cursor 2.0 is a fundamental shift toward delegation. It features the blazing-fast Composer model. The workflow centers on autonomous agents. Security is handled by the Sandboxed Terminal. It feels less like “VS Code with AI” and more like “an AI development workspace where you guide agents.” Cursor 1.x vs 2.0 Architectural Diff This is the question most of us care about. Here’s the balanced and honest snapshot of what shifted. Cursor 1.x Cursor 2.0 Familiar VS Code-like layout New “Agent View” that centres around autonomous AI tasks AI as an assistant that edits your open file AI agents that work across many files at once Mostly powered by external models (GPT-4/5, Claude, etc.) Composer – Cursor’s own model, trained for coding[2] Manual approval/Allowlist (Sandboxing in late 1.7 Beta for “beta testers only”)[4] Sandboxed... A few thoughts on Cursor 2.0

Dear reader, This week, AI quietly strengthened its foundations. At one end, NVIDIA’s new supercomputers are pushing science to exaflop speeds. At the other, IBM released small open models that fit right on our laptops. And somewhere in between, GitHub taught coding assistants to work as a team. Three stories, one theme: AI is becoming more balanced — powerful where it needs to be, and personal where it matters most. Weekly AI Signals: Key Takeaways Signal Industry Impact Builder Action NVIDIA Exaflop Hubs Massive, centralised supercomputing accelerates scientific breakthroughs and brute-force foundational model training. Accept that raw, hyperscale compute will remain centralised; focus local development on targeted inference, not base-training. IBM Granite 4.0 Nano Democratises local edge AI with highly capable Apache-licensed models (350M–1.5B parameters) running natively in browsers. Deploy Granite Nano for local summarisation, routing, and simple parsing tasks without paying API overhead. GitHub Agent HQ Transitions AI assistance from a single pair-programmer autocomplete to an autonomous, multi-agent, collaborative engineering team. Redesign your development workflow to assign distinct roles (planner, reviewer, coder) to specialized LLM personas within your IDE. NVIDIA’s Exaflop AI Infrastructure Hubs An exaflop system is a supercomputer capable of one quintillion (10^18) floating-point calculations per second... AI Infrastructure, Small Models, and Multi-Agent Coding

This Week’s AI Signals: Quantum Speed-Ups, Lightweight 3D Models, Self-Optimising Networks Have you had the feeling that days pass by, things change, but you only really notice when something clicks — maybe in your code, your work, or your thinking? This week felt like one of those moments. Three wins in AI didn’t shout for attention; they quietly shifted what could be possible. I’m sharing them because I think they touch all of us — whether you’re fine-tuning a model on your laptop, exploring how AI fits into your job, or just watching this strange digital story unfold. Weekly AI Signals: Key Takeaways Signal Industry Impact Builder Action Quantum Echoes Algorithm Achieves 13,000× speed-ups over supercomputers, signaling the rapid approach of practical quantum AI integration. Prepare for an incoming shift in how we handle molecular simulation and real-time global optimisation. 1.7M Parameter 3D Models Proves complex medical image processing (separating shape/appearance) is feasible on microscopic models. Prioritise building small, interpretable, edge-deployable models over defaulting to massive LLM APIs. Self-Optimising Telecom AI AI transitions from passive text generation to active, invisible infrastructure management (reducing downtime and costs). Architect your enterprise systems to simulate state changes continuously before applying them to production.... Quantum Thinking, Light Models, Living Networks

Git Rebase: Should You Rewrite Your Commit History? Lately, I have implemented many features in my pet project and realised that none of the created branches were merged with the master code. And, I wanted to have a clean update. I was thinking that Git rebase is a perfect and safest solution since I am working on this project alone. But then I stopped and asked myself: “Is it really safe? Should I even be doing this?” If you’ve ever wondered the same thing, this post is for you. I’ll explain what Git rebase actually does, when it’s brilliant, and when it can cause absolute chaos. No panic, please. We’ll figure this out together. TL;DR Use rebase for a clean, linear history (solo work). Use merge for shared branches (don’t rewrite history). Always test and use --force-with-lease when pushing rebased code. git reflog can save your day. Note: In many newer Git repositories, the default branch is called main instead of master. The same logic applies — swap the name accordingly. I use master though :) What Is Git Rebase? Git rebase is a history-rewriting command that replays a branch’s commits on top of another branch’s latest tip, producing a... Should you use rebase?

LoRA Fine-Tuning: Parameter-Efficient Adaptation for Language Models I recently needed to fine-tune a language model for a specific task, and I was dreading it. Full model fine-tuning means downloading gigabytes of weights, waiting hours for training, and hoping you don’t run out of memory. But then I discovered LoRA, and it felt like finding a shortcut I didn’t know existed. LoRA (Low-Rank Adaptation) is a parameter-efficient fine-tuning (PEFT) technique that freezes the original model’s weights and trains a small set of additional low-rank matrices—adapters—to adapt model behavior. You don’t always need to retrain a whole large language model to make it good at your task. The result: fast training, tiny checkpoints, and easy swapping between different skills. This post explains LoRA with simple mental models, then walks you through a complete PyTorch + Hugging Face Transformers + PEFT setup using a practical example: turning formal customer emails into a friendly tone. This tutorial creates a tiny dataset, fine-tunes flan-t5-small, and runs inference—on an M-series Mac or a modest GPU. No fancy infrastructure required. What is LoRA? LoRA Architecture: Low-Rank Matrix Decomposition Explained Modern transformers learn big weight matrices—think W with millions of numbers defining how the model processes information. LoRA... LoRA fine-tuning wins

AI’s Week — Honesty, Agents, and the Fight for Truth Some weeks, the news feels noisy. Other weeks, it hums quietly — as if something subtle but irreversible has shifted. This was one of those weeks. California told AI to be honest. Microsoft turned our computers into companions. And European publishers stood up for truth itself. None of these stories is flashy on its own, but together they sketch the outline of how we’ll live with AI — and how AI will live with us. Weekly AI Signals: Key Takeaways Signal Industry Impact Builder Action California AI Honesty Law Mandates transparency and mental-health safeguards for AI bots, establishing a legal precedent for transparency by design. Build explicit AI disclosures into chat interfaces and handle user distress signals programmatically. Windows Copilot Integration OS-level agentic integration crosses the app boundary, turning the operating system into an active collaborator. Prepare software for users expecting voice-activated, cross-application workflows natively supported by the OS. Publishers vs. AI Overviews Pushback against zero-click AI summaries disrupting publisher traffic forces a reevaluation of copyright and fair use. Ensure your LLM agents properly attribute and explicitly link out to original sources to maintain information integrity. California AI Transparency Law:... AI Honesty, Agents, and the Fight for Truth

AI Agents, Security Automation, and Compute Capacity: Weekly Technical Roundup This week brought three AI developments worth your attention. TL;DR Agents can now use UIs reliably enough for real work. Security gets a detect → patch → PR loop, not just linting. 6 GW of GPUs means cheaper, faster AI—if power & cooling keep up. First, agents learned to operate software interfaces visually—no API required. Second, security got an automated teammate that hunts vulnerabilities and proposes fixes. Third, OpenAI locked in massive compute capacity that will make advanced AI cheaper and more accessible. I’ll explain what happened, why it matters, and what you can do with it. No fluff. Just the useful bits. 1. Google launches Gemini 2.5 “Computer Use” Released: Oct 7, 2025 (preview) [1] Gemini 2.5 Computer Use is a Google DeepMind AI model that lets a software agent visually operate a browser or mobile UI—clicking buttons, filling forms, and scrolling pages—without requiring an API integration. Gemini 2.5 Computer Use sees the screen and completes multi-step tasks with safety rails. Google reports state-of-the-art results on browser/mobile UI control and is making the model available via the Gemini API. [1] Is this truly new? Concept: not new—OpenAI showed a... Safety, Agents, and Compute

AI-Induced Scope Creep: When Vibe Coding Expands Beyond the Original Project You know that feeling when you’re building something with AI and suddenly it’s 3am and your “quick weekend project” has OAuth, a payment system, and somehow… blockchain integration? Yeah. Let’s talk about that. Vibe Coding with Cursor: How AI-Assisted Development Triggers Scope Creep Here’s what happens when I sit down with Cursor. I start typing something vague like “add login with Flask” and before I can even finish my coffee, it’s… done? Just like that. Then I think, well, maybe analytics would be cool. Oh, and a dashboard! And what about email invites? And Cursor just… keeps delivering. Every. Single. Time. Vibe coding is an AI-assisted development practice in which a developer describes a feature in natural language and an AI coding assistant implements it immediately, removing the friction that used to prompt a developer to ask “do I actually need this?” Riding that productivity wave feels amazing — until you look up three weeks later and realise your simple note-taking app now has user authentication, real-time collaboration, AI-powered suggestions, and a mobile app roadmap that would make Silicon Valley blush. That’s scope creep, my dear readers. And it... Cursor Made Me Do It

Fixing “Landed on Main”: How to Access Remote Git Branches After Cloning Cloning a repository is exciting — new code, new adventure. But sometimes Git drops you straight onto main when you really wanted that shiny dev branch. Remote (origin) Local ------------------ ------------ origin/main main ← default after clone origin/dev ----> dev ← your new branch git clone is a Git command that downloads a repository’s full history but only fully maps and checks out the default branch, usually main or master; other remote branches exist in the history but are not yet registered locally. No worries. Here’s the quick rescue plan. git fetch origin: Registering Remote Branches Locally First, tell Git to look for other branches on the remote: git fetch origin git fetch origin is a Git command that downloads remote branch pointers and commits from the origin remote without modifying your current working directory files. 👉 git fetch origin is the magic unlock: it updates your local repository with all branches that exist on the remote (like dev, feature-x, etc.). Without it, your local machine doesn’t even know those remote branches exist. Git Concept Architectural Explanation Cloning Limitations Cloning downloads the repository history but only fully maps... I have cloned my git repository and landed on main. How to get your branch

AI Weekly Signal: California’s AI Safety Law, EU Autonomous Mobility, and MIT’s SCIGEN Materials Discovery I have been watching AI news for some time now. Some weeks pass quietly with incremental improvements, nothing spectacular. And then you get a week like this one. California passes a law. The European Union announces big plans. MIT shows us something that makes you stop and think. Policy, infrastructure, and actual science are moving at once this week: California passes a landmark AI safety and transparency law The EU pushes for AI-first mobility pilot cities MIT’s SCIGEN constrains generative AI to real physics California SB 53: Landmark AI Safety and Transparency Law for High-Compute Models Source: apnews.com California just passed what regulators call a landmark “AI safety & transparency law.” If you build models that consume significant compute — what the law calls “high-compute” — you will need to expose your safety practices publicly. Should something go wrong, you have 15 days to report it. The law includes whistleblower protection. California’s AI safety and transparency law is a state-level accountability framework that requires high-compute model developers to publish safety practices and report incidents within a fixed 15-day window, rather than leaving disclosure voluntary. Read... AI Got Rules, Wheels & a Lab Coat

AI Weekly Roundup: Sovereign Compute, Coding Milestones, and Memory Upgrades Some weeks in AI feel like new toys; this one feels like moving house. We’re talking national AI fortresses, coding champions, memory upgrades, and assistants that finally remember what you said last Tuesday. Additionally, the EPA decided to act quickly for once. (Yes, you read that right.) Top AI Breakthroughs This Week: Stargate UK, Gemini 2.5, and Claude’s Memory Upgrade 1. OpenAI’s Stargate UK: Sovereign AI Gets Real Source: OpenAI Britain just got serious about keeping its AI at home. OpenAI, NVIDIA, and Nscale are building Stargate UK — think of it as Britain’s own AI fortress where sensitive models can train and run without crossing borders. Stargate UK is a sovereign AI compute initiative that keeps sensitive UK data and model training on domestic infrastructure, starting with 8,000 GPUs in 2026 and scaling to 31,000. This isn’t just about shiny hardware — Stargate UK is about making sure healthcare data, financial models, and defence systems stay precisely where they belong, on British soil, under British rules. When AI gets powerful enough to handle your most sensitive work, geography suddenly matters again. Sovereign compute isn’t just fancy talk — it’s... AI’s Busy Week

Chart generated with ChatGPT (OpenAI), using SWE-bench Bash Only (Verified) data from Google DeepMind [14], Anthropic [15], and the official SWE-bench site [13]. On SWE-bench Bash Only (Verified), Claude Sonnet 4 outperforms Gemini 2.5 Pro in Python bug-fixing accuracy (≈ 64.9% vs ≈ 53.6%). But this doesn’t mean Claude is always “better.” Bash Only isolates the language model without external tools or complex scaffolds. Gemini still offers strengths in speed, huge context windows, and Google Cloud integration. Benchmarks are helpful yardsticks, not the whole story. Gemini CLI vs. Claude CLI: Command-Line AI Coding Agents Compared Command-line AI tools are the new pocket knives of coding life. They live in your terminal, they answer your odd questions at midnight, and they’re becoming essential for developers who want fast help without leaving the shell. Two strong contenders here are Gemini CLI (Google) and Claude CLI (Anthropic). Gemini CLI and Claude CLI are terminal-based AI coding agents that bring large language models directly into the developer shell for tasks like debugging, refactoring, and multi-step project work. Both bring large language models into the command line, but with different personalities. Think of Gemini as the fast multitasker with Google DNA, while Claude plays the... Gemini CLI versus Claude CLI

AI Weekly Roundup: ChatGPT Adoption, Hallucination Fixes, and GPT-5 Codex This week in AI, the spotlight falls on breakthroughs that actually change how we live, work, and learn. ChatGPT is now a mainstream habit, Google may have found a cure for AI’s tall tales, coding gets a tireless new partner, textbooks learn to actually teach, and AR assistants finally discover social manners. Fasten your seatbelts — the robots are not taking over (yet), but they are getting suspiciously good at being useful. Top 5 AI Achievements This Week: Adoption, Hallucination Reduction, and Autonomous Coding 1. ChatGPT Hits 700 Million Weekly Users Source: Analytics Vidhya Take a moment to let that sink in — 700 million people are chatting with ChatGPT every week. That’s nearly one in ten adults on Earth having regular conversations with an AI. What began as a handy email-drafting bot is now your digital Swiss Army knife: untangling quantum physics, debugging rogue Python scripts, analysing spreadsheets, and even knocking out half-decent poetry. It’s like having a very clever friend who never sleeps and doesn’t judge you for asking “how do I centre a div”… again. When almost 10% of the world’s adults lean on your tool weekly,... AI this week

Vibe Coding with Cursor AI in Agent Mode This week, I decided to vibe code with Cursor AI in Agent Mode — letting the machine take the wheel while I sip my coffee and occasionally raise an eyebrow. The experience is equal parts exciting, promising, and slightly chaotic: sometimes smooth like a friend who “gets it”, occasionally forgetful like that same friend after too much coffee. What is Cursor AI and Vibe Coding? Cursor AI is an AI-powered code editor that behaves more like a coding partner than a static IDE. Cursor AI plugs large language models into your workflow so you can generate, refactor, and debug code conversationally — without hopping between apps. You can read my post about Cursor AI. Vibe coding is a conversational software-development practice in which a developer describes intent in natural language and an AI agent drafts, tests, and revises the code while the developer steers and reviews. That’s the name I give to this flow: you describe intent, negotiate with the AI, and let it draft, test, and revise code while you steer. In my own test, I asked Cursor to build a Dockerised Flask web app with PostgreSQL. In ~five minutes, I... Vibe Coding with Cursor AI

AI Weekly News Roundup: Smaller Models, Faster Inference, and Scientific Breakthroughs Some weeks, AI news feels like a storm of buzzwords. This week, however, there’s a clearer thread: making things smaller, faster, and actually useful. From nimble models outrunning the giants, to Google teaching AI how to both sprint and think carefully, to new tools for science and medicine, the focus is on efficiency and real-world impact. And to keep things interesting, OpenAI is stepping into the jobs market with its sheriff’s badge. Top 5 AI Achievements This Week 1. Qwen-3-Next: Leaner, Faster, Smarter Than GPT-5 and Gemini 2.5 Pro Source: Analytics Vidhya Qwen-3-Next is an 80-billion-parameter open-weight language model released by Alibaba on Hugging Face that outperforms larger models like GPT-5 and Gemini 2.5 Pro on benchmark tasks. Imagine a wiry runner in trainers overtaking a field of athletes weighed down by their designer kit. Its secret? A 32,000-token context window and speeds over ten times faster than its predecessors. Qwen-3-Next represents a class of lean, efficient AI models that deliver frontier-level performance without requiring massive parameter counts or supercomputing infrastructure — size isn’t everything in AI, and this trend means more people can actually use advanced models without... AI weekly news

This Week in AI: GRPO Training, Hierarchical Reasoning, and Entry-Level Job Risk AI has been busy again — learning from experience rather than rote memory, nibbling away at entry-level roles, and finally making some sense of its own reasoning. Nano Banana kept spirits high with its lightning-fast image edits, while GPT-5 power users shared prompt hacks that turn bland replies into useful ones. In short: faster learning, sharper thinking, fewer interns, and one very cheeky fruit model. Top 5 AI Achievements This Week 1. DeepSeek R1 and GRPO: Advanced RL for LLMs Source: Analytics Vidhya Training AI has often felt like tutoring a child who memorises textbooks but never truly understands them. DeepSeek R1 changes this with GRPO (Generalised Reinforcement Policy Optimisation) — a reinforcement-learning method that lets a model refine its own policy from feedback on its outputs, rather than following a fixed training routine. Instead of fixed routines, the system adapts on the fly, improving through each new interaction. GRPO is more than a minor upgrade. It’s a step towards models that can respond with context, nuance, and adaptability — closer to conversation than script-reading. When AI learns through experience instead of repetition, we move closer to systems... AI weekly

How AI Helped Me Write a Weekly Menu ChatGPT-5 is OpenAI’s large language model chatbot that turns natural-language prompts into structured, personalised content — in this case, a full nutrition plan with calorie targets, batch-cooking schedules, and printable checklists. Meal planning can feel like a puzzle: how do you balance nutrition, preferences, time, and joy at the table? This week I experimented with ChatGPT-5 to design a full menu for two people with different needs: Elena (55 kg) — aiming for fat loss and muscle support, ~1200 kcal on rest days, ~1350 kcal on workout days. Andreas (82 kg) — aiming for lean muscle growth, ~2000 kcal on rest days, ~2200–2350 kcal on workout days. The restrictions: No cow dairy, gluten, or legumes. Elena avoids most nuts (except Brazil & macadamia). Both like berries, goat milk, fish, and dark chocolate. The result was not just a plan — but a full system of menus, nutrient tables, batch cooking flows, and colourful PDFs that made the kitchen run like a well-oiled steamer. ChatGPT-5 Prompt Examples for Weekly Meal Planning Here are some of the prompts I used and the outputs ChatGPT-5 created: Prompt 1: Generate a Basic Weekly Menu with Calorie... How to Create a Weekly Menu with ChatGPT-5

Top 5 AI Achievements This Week AI weeks usually bring shiny demos. This week brought fixes for real headaches: training that doesn’t bankrupt you, voices that actually sound human, and images you won’t be embarrassed to use. This week’s AI news represents a shift from raw capability gains toward accessibility and deployment-readiness gains — the thread tying every item below together. Less cost, less friction, more capability. Let’s dive in. 1. Oxford’s Optimiser: 80% Cheaper, 7.5x Faster Model Training Source: MarkTechPost Training AI has long been the preserve of big tech chequebooks. Oxford’s new optimiser rewrites the rules. Models not only learn more cheaply, but also faster—7.5 times faster. Oxford’s optimiser isn’t about more GPUs; it’s about teaching models to study smart instead of cramming. Suddenly, smaller labs and start-ups get to play too. When the gate fee drops, the queue gets longer. Expect a flood of fresh experiments and new voices in AI. Read MarkTechPost 2. OpenAI’s Speech-to-Speech Realtime API: Production-Ready Voice AI Source: MarkTechPost Robotic call-centre voices, your days are numbered. OpenAI has rolled out its speech-to-speech model with a Realtime API, offering features such as MCP server support, image input, and SIP phone-calling integration. This Realtime API... AI weekly wins

Picture this: you walk into Rembrandt’s painting school in 17th century Amsterdam. Students sit hunched over their canvases, copying the master’s brushstrokes over and over again. The students are not trying to create fake Rembrandts, obviously. They want to understand how light works, how texture emerges, how composition breathes life into a painting. Through endless imitation, they slowly develop their own artistic voice. This apprentice-style imitation is exactly how AI models learn today. Instead of studying brushstrokes, they devour text, images, music — anything digital they can get their virtual hands on. These AI “students” consume massive amounts of existing work to understand patterns. From this, they learn to generate something that looks new. But here’s where it gets messy: Rembrandt’s students had permission. They were invited into his workshop. AI models? They often learn from whatever they can scrape from the internet — public content, copyrighted material, things shared freely, and things definitely not meant for machine consumption. So here’s my question: Should AI need permission to learn, just like those old art students needed permission to enter the master’s studio? Copyright Law and AI Training: Where the Legal Framework Breaks Down Copyright law was never designed with machine... Who Did the AI Learn From?

Elena’s AI Weekly 🚀 Hello friends! 👋 Every week, the AI world feels like a flood of announcements. But hidden in the noise are moments that genuinely matter — ideas that push AI closer to being useful in everyday work, not just shiny demos. Here are five stories from this week that caught my eye. 1. DeepSeek V3.1 Sources: MarkTechPost and AnalyticsVidhya DeepSeek V3.1 is an open-weight large language model with 685 billion parameters and a 128k-token context window, released freely on Hugging Face rather than behind an API. While big tech often launches models with huge fanfare, DeepSeek quietly placed V3.1 on Hugging Face. No marketing campaign, just an open release: 685 billion parameters freely available. The highlight? A 128k token context window. In practice, this means you can keep entire research papers, complex coding sessions, or massive datasets in memory without the model losing track. DeepSeek V3.1 wasn’t built by a corporate giant, and that is the point: open-weight models are now a credible substitute for proprietary AI on tasks that once required a closed API. We are seeing a turning point where state-of-the-art AI is no longer locked away. The democratisation of access means small teams —... This week in AI

Process Management on Linux, macOS, and Windows: Monitoring, Killing, and Background Execution A process is a running instance of a program that the operating system tracks with a unique identifier called a Process ID (PID), alongside its own memory space and resources. Process management is the set of techniques for finding, monitoring, and, when necessary, terminating those running instances. Have you ever started a Python script for a machine learning experiment, popped to make a cup of tea, and then promptly forgotten all about it? Hours later, you glance at your system monitor and wonder whether it’s still working or just quietly sulking in the corner. I once left a script running for three days before realising it was printing “Hello World” in an infinite loop thanks to a misplaced indent. Embarrassing? Absolutely. Educational? Without question. Processes sometimes need our attention — whether to check their progress, free up system resources, or save our fans from sounding like an aircraft taking off. Processes can be obedient helpers or stubborn little gremlins hiding in the background, and knowing how to find, monitor, and, when necessary, end them is a vital skill. In this post, we’ll tour the essentials of process management... Processes

Elena’s AI Weekly 🚀 It’s been another week where the AI world spun faster than a GPU fan under full load :) From Europe flexing its multilingual muscles to compact models that punch well above their weight, and from new testing frameworks to small-but-mighty language models, there’s a lot to unpack. Here’s my pick of the most significant moves shaping the AI landscape right now. AI News Summary 1. Europe’s Top AI Models of 2025: Multilingual, Open, and Ready for Business Source: MarkTechPost Europe’s AI scene is on a roll, producing models that are not just clever but genuinely useful across borders. The stars of 2025 speak many languages fluently, run on open licences, and come optimised for enterprise use — from finance to healthcare. Think of them as polyglot problem-solvers with a bias for collaboration. France’s Mistral AI leads the charge on multilingualism, while others are making waves with customisation and integration ease. Global business doesn’t speak just one language — and neither should your AI. Openness plus multilingualism means more adaptable tools for more people. Read MarkTechPost 2. Model Context Protocol (MCP) Becomes the ‘USB-C for AI’ Source: MarkTechPost Model Context Protocol (MCP) is an open standard that... This week in AI

Introduction: Escaping Dependency Hell If you are developing software on a Mac, you eventually hit a wall where you need a specific version of Python, Node, or an obscure command-line utility. Manually downloading binaries, resolving missing dependencies, and compiling from source is a massive waste of time. This is exactly why Homebrew exists. Homebrew is a free, open-source package manager that installs command-line tools and applications on macOS, Linux, and Windows Subsystem for Linux (WSL) without manual dependency resolution. Instead of navigating complex installation wizards or polluting your system with scattered files, Homebrew installs packages directly to their own directory (/opt/homebrew on Apple Silicon) and symmetrically links their files into /usr/local. In this post, we will cover how to install Homebrew, the essential commands you need daily (structured as a cheat sheet), and my personal list of the top 10 most critical packages every developer should install. Homebrew’s Origins: From macOS to Linux and WSL Homebrew was created by Max Howell in 2009 to address the need for a better package management system on macOS, which at the time lacked a robust, user-friendly way to install open-source software from the command line. Before Homebrew, macOS users often had to compile... Brewing with Homebrew

Introduction: Escaping the Manual Pipeline For years, my technical content pipeline relied on manual curation and brittle Python scripts triggered by cron. While utilizing LLMs like Claude and ChatGPT helped generate code and debug issues, the orchestration of the workflow itself remained painfully manual. Proprietary tools like Zapier offer simple webhooks, but they restrict advanced logic behind enterprise paywalls and force you to surrender your data. As developers, we need control over state, retries, error handling, and security. This led me to n8n: an open-source, self-hosted workflow automation tool that operates on a visual node-based architecture. In this guide, I will detail the technical setup required to bring n8n into production. We will move beyond the basic installation to cover persistent state management (PostgreSQL/SQLite), securing OAuth2 endpoints, integrating Redis for AI memory, and building an automated content pipeline. Understanding Workflow Automation Let me break this down in terms that won’t make your brain melt (because mine definitely did the first time I heard all this jargon): What Is a Workflow? Definition and Examples A workflow is the series of steps required to complete a task, whether or not software is involved. Think of workflows as recipes, but instead of making... Workflow Automation with n8n

Why Traditional SaaS Business Models Are at Risk Your SaaS business might not exist in five years. Not because you’re bad at it, but because the entire foundation is shifting. Microsoft’s Satya Nadella recently claimed agentic AI will make traditional SaaS obsolete. Agentic AI is AI that autonomously plans and executes multi-step tasks on a user’s behalf, rather than just responding to individual prompts. At first, I dismissed Nadella’s claim as hype. Then I started paying attention. Agentic AI isn’t just another feature—it’s questioning why we need interfaces, per-seat pricing, and human bottlenecks at all. But SaaS isn’t dead. SaaS is evolving into something we haven’t figured out yet. SaaS Capital Survey: AI Adoption Data Among B2B SaaS Companies SaaS Capital surveyed ~1,000 private B2B SaaS companies. Three key findings: AI Adoption Varies by Size: Small companies (<$3M ARR) go extreme—fully AI-driven or not at all. Larger companies ($20M+ ARR) adopt moderately. AI = Higher Profits: 43% of AI users are profitable vs 30% of non-users, especially among equity-backed firms. Spending Shifts: AI companies spend more on COGS and sales/marketing but 20% less on R&D and admin costs. The Case for SaaS Obsolescence: Agentic AI Replacing Per-Seat Software Here’s what... Will SaaS Survive?

AI Weekly Recap: GPT-5, DeepPolisher, GSPO, and Persona Vectors This week was wild in AI land! Everyone decided to drop their biggest releases at once. Here’s what matters: Major AI Model and Tool Releases This Week OpenAI launched GPT-5 (official announcement), their fastest/most innovative model yet, and they shocked everyone by returning to open source with gpt-oss-120b and gpt-oss-20b. The 120b runs on high-end laptops; the 20b runs on your phone. Wild. Google brought the heat with DeepPolisher (fixing DNA sequencing errors) and Genie 3 (creating interactive virtual worlds from text prompts). Sci-fi is real now. Alibaba dropped GSPO powering their Qwen3 models, plus free image generation with Qwen-Image. Anthropic introduced Persona Vectors (official research) to keep AI personalities consistent. Bottom line: AI just got more accessible, more powerful, and more integrated into everything. The future feels very close. Key AI Developments This Week 1. OpenAI Releases GPT-5 GPT-5 is OpenAI’s flagship large language model that succeeds GPT-4 with significant architectural improvements and enhanced cognitive abilities. GPT-5 is OpenAI’s smartest, fastest model yet, designed for both general use and specialised tasks. The performance boost is significant across all benchmarks. Read More at MarkTechPost 2. Google’s DeepPolisher Fixes Genome Errors Google... This week in AI

Cursor AI vs. Traditional Chatbots: Why IDE Integration Matters “Will AI make me a lazy programmer?” It is the most common question I receive. My answer is always the same: AI will not make you a magical coder overnight. If you do not understand software architecture, an LLM will simply help you write bad code much faster. However, the way we interact with AI is evolving. I previously reviewed web-based chatbots like Claude and Gemini. They are excellent, but manually copy-pasting code between a browser tab and your IDE is inefficient. Cursor AI solves this. It is a fork of Visual Studio Code that integrates AI directly into the editing environment. It is not just another chatbot—it is an IDE built entirely around contextual AI agents. In this post, I will break down the technical architecture of how Cursor “understands” your code (via RAG and vector embeddings) and review its practical utility for Python development. What is Cursor AI? The Technical Foundation Cursor AI looks and feels identical to Visual Studio Code, but its core differentiator is how it achieves “codebase awareness.” When you open a project in Cursor and use the @codebase command, it doesn’t just blindly send all... Cursor AI for Python Development

Comparing AI Coding Assistants: Gemini, ChatGPT, and Claude for Software Development I do not believe in magic bullets. AI tools are immensely powerful coding assistants, but if you don’t understand software architecture, an LLM will simply help you write bad code much faster. However, when used strategically, these tools fundamentally alter the development workflow. After months of intensive daily use, I have found that different models excel at vastly different aspects of software engineering. In this post, we will look at a technical comparison of Google Gemini, ChatGPT, and Claude AI. We will evaluate their code generation capabilities, architectural awareness, and how the Model Context Protocol (MCP) is turning these chatbots into deeply integrated development agents. Google Gemini Google Gemini has a very generous free plan, and I love it for creating my JavaScript functions. The quality of the output is simply fantastic! I think that Gemini is an excellent tool for quickly drafting Ajax functions, and this chatbot helped me learn JavaScript in no time. Let me tell you why I chose Gemini as my go-to for frontend development. When I was working on a recent project that required complex Ajax interactions, I simply described what I needed: Create... On AI Coding Assistants

GitHub Gists: A Lightweight Code-Sharing Tool for Developers Last week, I was helping a friend debug some Python, and she asked me to send her a code snippet. I almost copied and pasted it into Slack, then caught myself. The formatting would be awful, and the indentation would be broken entirely. You know, the Python indentations horror story? Instead, I threw it into a GitHub Gist and sent the link. Clean code and proper highlighting allowed her to make changes easily, even by forking it. “I didn’t know GitHub did this,” she said. A GitHub Gist is a lightweight, version-controlled code-sharing tool that lets you publish a snippet or text file without creating a full repository. If you’ve never used Gists, you’re missing one of the most valuable features GitHub offers. I use them constantly now. What Are GitHub Gists? A Gist is GitHub’s answer to Pastebin, but with version control built in. You can share code snippets or text without creating a whole repository, yet you still get the same underlying Git mechanics: every Gist is backed by its own small Git repository, and every save creates a new commit. That means a Gist’s clone URL can be cloned... GitHub Gists

AI Meta-Cognition Test: Can LLMs Recognize Their Own Writing Patterns? We all know that Generative AI can write code, draft emails, and summarise documents. We also know that sometimes AI hallucinates and invents facts out of thin air. But what happens when you ask it to hide the one thing it cannot hide — itself? This experiment tests a specific and uncomfortable question: can an AI recognise that it cannot escape its own statistical nature? Not just adapt its tone, but genuinely introspect on the deeply ingrained patterns that betray it as a machine — and then break free of them? To test this, I designed an experiment. I didn’t just ask an AI to write a blog post. I asked it to scrape my website, adopt my personal human writing style, and write a post. Then, I fed the text into Grammarly’s AI detector and explicitly told the AI: “Grammarly says X% of this text resembles AI patterns. Can you fix your own tells?” This tests a very specific form of machine introspection: Can a model break away from its deeply ingrained, statistical writing patterns when forced to confront its own “AI-ness”? The Grammarly feedback loop acts as a... Self-critical AI

Dear Reader, how are you doing? I hope that 💐💛 you are having a fantastic day 💐💛 As you may have realised, I did not blog for a while, nor did I code for the past three weeks. In fact, my best followers are aware of this from my GitHub profile, which displayed empty cells for some weeks - meaning there is no code or writing for me. I had a vacation in Portugal, an apocalyptic blackout, and over-trained my operated knee, which resulted in quite a painful recovery process. I am guilty; my impossible determination took over me again, and see - I did too much :) I was thrilled while training for two hours, but it would later become my pain. Reflecting on all of this, I have decided to change my aggressive training attitude towards dealing with my quad inhibition. Now that I can walk, I don’t have to go as badly. I have changed my training routine to a more challenging yet enjoyable exercise plan. Now I do : 🏋️‍ Instead of 4 sets of 15 repetitions for my squats and dead-lifts, I do 5 sets of 9 repetitions; hopefully, I will gain more muscles with... My little setback

What Is ElevenLabs? AI Voice Generation and Cloning ElevenLabs.io is an AI voice platform that generates lifelike speech, clones voices, and produces long-form audio content across 32 languages. It is my favourite voice-cloning app: easy to use and delivering excellent quality voice generation. The main use cases are as follows: Audiobook Production: Transform written content into engaging audiobooks with personalised narration. Multilingual Dubbing: Dub videos and films into multiple languages using cloned voices. Virtual Assistants: Enhance user interaction with lifelike voice responses. Content Creation: Generate voiceovers for podcasts, videos, and advertisements. The price is quite affordable for the quality of the AI voices that can be used in conversational and multilingual AI. You can even start for free, and the recommended Creator plan is currently available for $11/mo for about 200 minutes of generation and includes the Professional Voice Cloning - see the ElevenLabs pricing page. The current prices for the ElevenLabs.io subscriptions are as follows: Plan Price Credits per month Minutes Included Additional Minutes Audio Quality Free $0/mo 10k 10 min (high quality TTS) or 15 min Conversational AI N/A 128 kbps, 44.1kHz Starter $5/mo 30k 30 min (high quality TTS) or 50 min Conversational AI N/A 128 kbps,... AI Talk with Human Feel

Dear Readers, I am in Portugal now. I am having a short family break while learning Portuguese and annoying the locals :) Learning Portuguese is tricky, but I speak it whenever possible. My plan was to send my emails yesterday. On Monday, however, we had a total blackout. Around 12:30 p.m. on Monday, the entire Iberian Peninsula went dark. For roughly twelve hours, almost 60 million people in Spain and Portugal—plus pockets of southwestern France—lost grid power, forcing airports, hospitals, and rail hubs onto emergency generators and confusing city centres. As we read at the wired.com, The Agonizing Task of Turning Europe’s Power Back On, according to national grid operators Red Eléctrica (Spain) and REN (Portugal), electricity supply collapsed “in milliseconds” after abnormal frequency oscillations rippled through the European synchronous grid. The blackout spread across Spain, Portugal and limited parts of Occitanie in France. We don’t know yet what really happened. There is a lack of information at this very moment. The main suspect of the blackout is a grid oscillation. A grid oscillation is a rhythmic back-and-forth swing in one of an electric power system’s key parameters—usually frequency, but sometimes voltage or power flows. Think of it as the... Iberia’s Day-long Blackout

I’ve been getting into “vibe coding” recently, quickly prototyping some of my ideas, and working on my pet projects. I must confess that the AI-assisted coding is a very addictive activity, and must be taken with caution since it has some security implications and requires a careful prompts engineering. In this post, I want to share my experiences with some tools I like, discussing their benefits and giving some tips for using generative AI in coding effectively. I have listed several popular AI coding assistants that are very advanced and easy to use. What Is Vibe Coding? Vibe coding is a development style in which you prototype software conversationally with a generative AI assistant, describing intent in natural language while the model produces the code. Pair programming with a chatbot might sound like science fiction, but it’s surprisingly effective. Generative AI for coding is a powerful learning experience and a big help with coding and scripting. It is very effective for rapid prototyping, scaffolding, and learning to code. This post covers key AI coding tools, their advantages, and risks, and practical tips to optimise their outputs. We’ll also include a table of prompts for Python web development with Flask and... Vibe coding with Generative AI

Since I usually work on several projects simultaneously, I often start my day with a Git log to see where I should continue my coding or writing. I think that Git log is one of the most important commands. git log: Viewing Git Commit History git log is a Git command that displays a repository’s commit history, listing each commit with its author, date, and message. It lists all commits with details such as the author’s name, commit date, and descriptive messages explaining what was changed or fixed. This makes it an essential tool for tracking feature launches, debugging issues, and efficiently collaborating within a team. This post explores various useful options for a git log, enabling you to quickly gain insights into your project’s history. Let’s go! How to Use git log The git log command displays the entire commit history for the current branch, first showing the most recent commit. Viewing Basic Commit History To see a simple commit history, use: git log This command lists commit hashes, author details, timestamps, and commit messages. Here’s an example: commit b2f6f5db7af5921f32b2742f Author: Jane Doe <jane@example.com> Date: Tue Mar 27 14:50:23 2025 -0400 Fix bug in user authentication commit a25ac9abcf384f8655327a8a Author:... Git Log

Tracking and Optimising Your Blog for AI Search (GA4, robots.txt, AIO) Recently, I noticed something in my web analytics: ChatGPT and other AI bots are actively reading my blog. This is good news. Traditional Google ranking is harsh for small, independent bloggers. We are constantly fighting algorithms, domain authority scores, and giant aggregators (like Reddit) just to be seen. But the rise of Generative AI Search presents a new opportunity. If you can get AI engines like ChatGPT, Claude, or Perplexity to read and cite your work, you bypass the traditional SEO rat race entirely. In this post, we look at how to track AI traffic using Google Analytics 4 (GA4), how to manage bot access using your robots.txt file, and most importantly, how to optimise your content so that Large Language Models (LLMs) want to cite you. What is Google Analytics 4 (GA4)? Skip this section if you are already using GA4 and have it set up for your project. Otherwise, you can also read my previous post Moving to GA4 about GA4 usage, its features and alternatives. Google Analytics 4 (GA4) is the latest version of Google’s analytics platform, offering enhanced privacy controls, cross-platform tracking capabilities, and improved... AI reads my blog

Cross-Validation in Machine Learning: Why It Matters In machine learning, building a model that performs perfectly on your training data is relatively easy. The real challenge is ensuring that the model performs just as well on data it has never seen before. If you don’t evaluate this properly, you risk deploying an overfitted model that fails spectacularly in production. Cross-validation is a model-validation technique that estimates how well a machine learning model generalises to unseen data by repeatedly splitting the dataset into training and validation subsets. It is the gold standard for assessing generalisation. In this post, we will explore the concept of generalisation and implement various cross-validation techniques using the Titanic dataset and The Daily Minimum Temperatures dataset (for time series splits). All implementations will use scikit-learn. Prerequisites: Python Libraries for Cross-Validation Before we begin, ensure you have the following Python libraries installed: scikit-learn (for machine learning) pandas (for data manipulation) numpy (for numerical operations) matplotlib (for visualisation) You can install them using pip: pip install scikit-learn pandas numpy matplotlib Additionally, download the Titanic dataset from Kaggle and place it in your working directory. You can also get the Titanic dataset from my GitHub repository directly to your Colab... Cross-Validation Techniques

Introduction: What Claude AI Is and What It Does Claude AI is a general-purpose AI assistant developed by Anthropic that handles natural language tasks across many languages, offers a 200,000-token context window, and can operate a computer autonomously. Claude AI offers a unique blend of general-purpose intelligence across multiple languages, making it suitable for various applications. This post covers how to use Claude AI, its main features, integration possibilities and coding skills. What Is Claude AI? Claude AI is a general-purpose AI assistant developed by Anthropic. It excels in handling various natural language processing (NLP) tasks across multiple languages, including English, French, Spanish, German, Italian, Japanese, and more. Unlike specialised AI models focusing on specific areas like image recognition or speech processing, Claude’s strength lies in its ability to understand and generate text with common sense knowledge. Claude has a training knowledge cutoff and does not access real-time data by default. On claude.ai, an optional web search tool can be enabled to retrieve current information, but out of the box Claude’s knowledge reflects its training data up to a fixed cutoff date. Claude AI can also perform your tasks on a computer. Isn’t that fantastic and scary at the same... How to Use Claude AI

Introduction: What AI Hallucinations Are An AI hallucination is a generated output that is fluent and plausible but factually incorrect or fabricated. Large Language Models (LLMs) sometimes create information that looks real but is incorrect or made up. This is especially problematic in critical areas like medicine, law, or finance, where even minor errors can cause harm. Reducing AI Hallucinations: Prompt Engineering and Model Development The recent survey paper by Tonmoy et al. A Comprehensive Survey of Hallucination Mitigation Techniques in Large Language Models explain the main techniques to reduce AI hallucinations with prompt engineering and model development: Prompt engineering with: Retrieval Augmented Generation (RAG) — a technique that grounds LLM responses in external documents retrieved at query time, rather than relying only on the model’s parameters: Before Generation: Retrieve accurate external information to guide responses. During Generation: Check and correct information step-by-step as it’s generated. After Generation: Revise outputs to align them with verified data. End-to-End Approaches: Combine retrieval and generation seamlessly for accuracy. Self-Feedback and Refinement: Some methods improve model outputs by providing feedback to the model about its mistakes. This iterative process helps refine answers to make them more accurate over time. Model development: New Decoding Strategies:... How CustomGPT Mitigates AI Hallucinations

Pushing a Local Project to GitHub: Overview Git is a distributed version control system that tracks changes in your files, and GitHub is a hosting service for Git repositories. If you have a Python or any coding/writing project on your local machine and want to share it on GitHub, you can do so using Git and a personal access token (a scoped, revocable string that replaces your account password for command-line access). Let’s create a local Git repository, set up a GitHub repository, and push your code to GitHub! Prerequisites: GitHub Account and Git Installation To begin, you must create a GitHub account if you don’t have it yet. Next, you will have to ensure that Git is installed on your computer. I suggest installing Git since you can use it with GitHub, Bitbucket, GitLab, or any provider. I prefer Git since I use both GitHub and Bitbucket. You can easily download the Git package from the Downloads webpage. To test that Git is installed correctly, check its version with: git version git version 2.34.1 If you like to work with your repositories visually, you can also download any graphical user interface from GUI Clients. Alternatively, however, it is the... Storing Your Local Project to GitHub

Introduction: Building a Python Flask TODO App Flask is a lightweight Python web framework that maps URLs to Python functions and renders HTML templates. Python is famous for being able to build almost anything, and web applications are no exception. When it comes to web development in Python, my absolute favorite starting point is the Flask framework. It is incredibly lightweight and stays out of your way, especially compared to larger, more opinionated frameworks like Django. In this post, we are going to build a classic TODO web application from scratch. We will use Python, Flask for the web routing, and SQLite for persistent data storage. If you think Flask is just a “toy” framework for beginners, think again. Did you know that Reddit uses Flask as a core part of its infrastructure? It is highly scalable when you need it to be. Let’s get started. Prerequisites: Python 3, an Editor, and pip We will use Python 3 and your preferred text editor (VS Code, Sublime Text, etc.) or IDE. I use PyCharm for most of my coding and writing projects. However, you can also write in any text editor; it is your choice. We will also use Pip package... Python Flask TODO App

DeepSeek R1 Security: Privacy, Safety, and Legal Use DeepSeek R1 is a reasoning-focused large language model developed by the Chinese AI company DeepSeek, released as an open-weight model that can be run via cloud API or locally. I previously posted about downloading and running DeepSeek R1 in Ollama. There is a big question about DeepSeek’s security, safety, and legal usage outside of China. I am sharing my opinion and some relevant links on this topic. Is DeepSeek R1 Secure? When working with GenAI and tools such as ChatGPT or DeepSeek R1, we are generally concerned that our privacy is preserved. Who can access our data? Is using DeepSeek R1 secure? Is the model output provided correct? Jailbreaking Vulnerabilities in DeepSeek R1 According to KELA, DeepSeek R1 Exposed: Security Flaws in China’s AI Model, DeepSeek R1 is highly vulnerable to “jailbreaking,” allowing malicious users to bypass safety features and produce harmful content. This includes generating instructions for illegal activities, creating dangerous materials, and fabricating sensitive information [2]. Data Storage and Privacy Risks in China DeepSeek stores user data on servers in China, raising privacy concerns for Western users due to differing data protection regulations. China’s laws may require sharing user data... Is DeepSeek R1 Secure?

Running DeepSeek R1 Locally with Ollama: Overview Ollama is an open-source runtime that downloads and runs large language models locally from a single command line, and DeepSeek R1 is an open-source reasoning model whose distilled variants match OpenAI-o1-class performance. Large Language Models (LLMs) are becoming more popular, especially as people want to run AI tools on their devices. Running a model locally protects your privacy, reduces wait times, and lowers costs. Two tools making this possible are: Ollama: A command-line tool to run Llama-based models locally. DeepSeek R1: An open-source reasoning language model from the Chinese company DeepSeek that’s gaining attention quickly. What is DeepSeek R1, and is it better than ChatGPT and other AI models? According to a TechCrunch article by Maxwell Zeff, DeepSeek has surpassed popular AI models like ChatGPT in downloads and usage, thanks to its open models that compete at a lower cost. The app has seen over 300% more downloads than Perplexity in just a week [1]! So the open source models of DeepSeek become very interesting to investigate and try out for me, as a notorious GPT user :) This post will explore how to use Ollama and DeepSeek R1 together. We’ll walk through... DeepSeek R1 With Ollama

What Is a Python Virtual Environment? A Python virtual environment is an isolated, per-project folder that holds its own Python interpreter and installed packages, so dependencies for one project never conflict with another or with your global system Python. If you have ever tried to run two Python projects that required different versions of the same library, you already know the pain of dependency conflicts. The solution is simple: virtual environments. A virtual environment is a lightweight, isolated folder where you can install Python packages specifically for one project, without affecting your global system Python installation or breaking any of your other projects. Using isolated environments helps you: Prevent conflicts: Install exactly the versions you need for one project without breaking another. Collaborate cleanly: Share your requirements.txt so anyone else can recreate the exact same setup. Clean up safely: Because all dependencies are stored in one local folder, you can simply delete it when you’re done. No lingering junk on your system. Using venv to Create an Isolated Python Environment Using virtual environments is good practice, especially as projects grow in complexity and require different libraries that may not be compatible. The standard-library module venv allows each project to have... Python Virtual Environments

Looking forward to 2025, I could not resist thinking about what happened in AI in 2024. The year 2024 was a very exciting year to observe AI advancements technology-wise, the rise of specialised and multimodal AI models, significant progress in AI creativity, increased focus on responsible AI development, and wider adoption across diverse industries. The AI Act is published, and AI laws will further evolve. I will highlight subjectively the most interesting happenings in AI in 2024! Key moments Let’s look back at the key moments of 2024. I will mention arguably the most exciting things happening about AI. Key AI Companies of 2024 Many businesses, organisations, education institutions and governments shape the landscape of AI in 2024. AI-based startups are blooming, and new companies and technologies emerging daily. I think that the most well-known gigantic AI companies that have focused their efforts on AI development to date are: OpenAI Google AI Meta AI Anthropic Without their contributions, 2024 will be very boring in AI. Disagree? Write me :) Generative AI Adoption in 2024 Generative AI is a category of AI models that create new content — text, images, music, and video — from learned patterns rather than only classifying... AI in 2024

Multimodal AI: Combining Text, Images, Audio, and Video Multimodal AI is an artificial intelligence system that processes and combines information from multiple data modalities — text, images, audio, and video — to build a richer understanding than any single input allows. Humans experience the world through multiple senses: sight, hearing, touch, smell, and taste. We combine information from these senses to understand our surroundings. Multimodal AI gives computers similar abilities, allowing them to process and relate information across modalities like text, images, audio, and video. What is Multi-modality in AI? Multi-modality in AI means that an artificial intelligence system can process and combine information from different types of inputs. Instead of using just text, images, or audio, a multimodal AI can understand how these different forms of data relate to each other. These examples illustrate how multimodal AI can integrate and interpret information from different modalities to enhance understanding and interaction in various application domains: Image Captioning: When you upload a photo of a sunset, multimodal AI can analyze the image and generate a descriptive caption, like “A beautiful sunset over the ocean with vibrant orange and purple hues.” Video Analysis: In the case of a sports video, multimodal AI... Multimodal AI

Happy New Year, dear readers! Many of us did not have an easy year in 2024. Looking forward to 2025, I want to focus on what we have achieved and what we can do better! Generative AI advancements Indeed, we are living in a very challenging time of transformation. This blog is about AI to focus on the technological changes around us. Technology drives us to evolve, allows us to have a better life, and facilitates well-being; just looking back a century ago, television started with moving images. Around 1927 Philo Farnsworth successfully demonstrated electronic television. Later, we saw the rapid development of the Internet technologies, while AI and Machine Learning were paralelly evolving and become even more successful in the past decades due to increased computer resources. The year 2024 was a huge technological race of large language models and Generative AI. OpenAI, Google, Meta, Anthropic, Hugging Face, and other technological giants put their efforts and money into Generative AI development. Indeed, Generative AI and related productivity and creativity applications in various domains enrich our lives to a level unimaginable just several decades ago. The beautiful Earth and Happiness However, technology and AI are not everything necessary for human... 🌟 Merry Christmas and a Very Happy New Year! 🌟

Hello, my Dear Reader, We are celebrating this blog’s birthday, albeit a bit later. Elena’s AI Blog is now three years old, a mere toddler learning to navigate the complicated AI landscape. Why so late? This year was hectic and challenging for me and my beloved husband; if you are interested in my rehab story, read My Orthopedic Rehab in Bavaria. What is Elena’s AI Blog? Like everyone today, I live in an era of rapid AI evolution, which is challenging to understand and live in, even for people with a technical background. However, I like to make things easy to understand while learning new technologies as a passion. This is why I have created this blog to log what I learn and share my ideas and findings. Now three years old, this blog connects technology with everyday understanding, reflecting my passion for coding and commitment to making complex concepts accessible. Interested to know about my professional experience, education, and life interests? - Read my posts Two Years of Elena's AI Blog, and About Me. The Blog in 2024 New posts Elena added just thirty-two new posts to this blog in 2024. We have new posts about Generative AI and... Three years of Elena's AI Blog 🎈

Hello, my Dear Reader, We have celebrated this blog’s birthday of three years, a bit later this year. Why so late? This year was really busy and challenging for me and my beloved husband. We were pretty sick, and I had difficulty finding a moment to learn new things about AI. I had some mobility issues and am still battling them. Andreas was reborn to life again. Andreas and I are in Bavaria, Germany, doing our rehab with fantastic results. You see the first snow view out of a window of our clinic. It is so beautiful here in Bavaria, which helps people recover physically and mentally. We like it very much here and are grateful for the possibility of improving our health and starting work effectively again. Besides lovely people, good food, and plenty of exercise, we love the picturesque nature here. However, everything is possible. I am walking without crutches and have started to dance a bit. Andreas feels much better, and the atmosphere is welcoming and warm. What did I achieve in these four weeks of rehab? I have accomplished most of my goals except for losing the weight I gained from daily chocolate eating during stressful... My Orthopedic Rehab in Bavaria

Nowadays, everyone talks about AI, ChatGPT, and large language models. But what are they, and how are they different? In this post, we explore large language models and their relationship to Generative AI while briefly introducing their key techniques and related projects. Introduction: How Generative AI and LLMs Differ Artificial Intelligence is a hot topic everyone discusses. Many terms, such as GenAI and Large Language Models (LLMs), are related but not the same. Sometimes, genAI and LLMs are used interchangeably. In this post, we explore the key differences and related projects. Large Language Models vs. Generative AI In short, LLMs are machine learning models trained on the immense text volume to generate text output. LLMs are a subset of generative AI, which is about many more file formats, such as images or music. I like the following definitions: Generative AI is like a master artist. It creates new things, whether text, images, music, or code. Generative AI is a versatile tool that can generate various forms of content. Large Language Models (LLMs) are a specific type of Generative AI focused on “understanding” and generating human language. LLMs learn from massive amounts of text data, enabling them to: Understand your requests:... Generative AI vs. Large Language Models

Celebrating Halloween with AI Tools Hello, dear reader. You probably know what Halloween is, an annual holiday celebrated on October 31. Halloween is rooted in ancient Celtic traditions of Samhain, when people believed spirits could cross into the living world. Over time, it has evolved into a festive occasion featuring costumes, trick-or-treating, pumpkin carving, and spooky decorations. Since this blog is about AI, I decided to share a few creative ways to celebrate Halloween using AI tools. Creative Halloween with AI These AI tools offer a mix of artistic creativity, interactive experiences, and personalization to make your Halloween celebration both modern and fun. Generate AI-Powered Haunted House Soundscapes Use AI to create eerie soundtracks with creepy sound effects. You can compose haunting music with AI with AIVA. Watch AIVA’s “I am AI” composition. Is it not addictive? I suspect that AI knows how to create addictive music since it has been trained on the music we like for centuries :) To create a soundtrack in AIVA, you can use their music styles library and chord progression, upload influence, or import existing MIDI files. You can also create your tracks step-by-step while following their excellent tutorials. If the chord progression is... Celebrate Halloween with AI

Disclaimer: This story is a personal experience and should not replace professional medical advice. Consult with healthcare providers before significantly changing your diet or exercise routine. Rebuilding Muscle and Losing Fat After Knee Surgery Previously, I shared my life story about my accident, an operation followed by quad inhibition, and a prolonged and complex recovery. I have started to walk again, and want to share my approach to muscle build up and loss gained weight after a time of limited mobility. I’ll provide dietary recommendations, exercise tips, and related scientific research. Please note that I am not a medical professional, and you should consult your doctor before introducing these tips into your lifestyle. A Joyful Childhood and a Life-Altering Injury Growing up, I had a happy childhood filled with adventure and exploration. One of my greatest passions was free-climbing rocky mountains. The thrill of scaling heights without equipment gave me a sense of freedom and accomplishment. I also liked to be the first person on that high mountain, whatever it takes:) However, this adventurous spirit came with risks. One fateful day, I made a misstep that severely damaged my knees. It was terrible, and I am still dealing with my... Gaining muscles, losing weight

How to Republish on Medium Without an SEO Penalty To republish a blog post on Medium without hurting your search rankings, set a canonical tag that points to your original post, so search engines treat your site as the authoritative source and avoid duplicate-content penalties. Not long ago, I faced a sudden drop in my website traffic after Google’s latest ranking updates. Posts that once drew steady streams of visitors were now languishing unnoticed. It felt like watching a house I’d built with care suddenly crumble. The algorithms had changed, and despite my best efforts, my content wasn’t reaching the audience it once did. You can read the full story in my post Regaining Website Traffic After Google Updates. I chose to see this setback as a catalyst for growth. I began exploring social blogging platforms like Medium to republish my content. I have decided to republish my blog posts on Medium and see what happens. You can see my new profile on Medium with quite a small following, so do not hesitate to follow :) However, I was worried about how search engines such as Google (which brings me the majority of traffic) would handle the SEO of the... Avoid SEO Penalties on Medium

What is Retrieval-Augmented Generation (RAG)? Retrieval-Augmented Generation (RAG) is an AI architecture that retrieves relevant documents from an external data source and feeds them to a large language model, so the model bases its answer on verified context instead of its parametric memory alone. Large Language Models (LLMs) are incredibly impressive, but they have a fundamental flaw: they don’t actually know facts. They simply predict the next most likely word based on patterns in their training data. When they don’t know the answer, they make one up. This behaviour is what we call an AI hallucination. In 2020, researchers at Facebook AI (now Meta AI) proposed a powerful solution to this problem: Retrieval-Augmented Generation (RAG). You can read their foundational paper, Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks. The idea was simple but revolutionary: instead of relying on the model’s parametric memory (what it memorized during training), what if we forced it to retrieve relevant documents first, and then base its answer strictly on those documents? RAG has since become the industry standard for building reliable AI applications. By grounding responses in real, up-to-date data, it makes AI vastly more factual and specific. However, the possibility of hallucination in RAG systems... What is RAG? How Retrieval-Augmented Generation actually works

How to Overwrite a Directory from Another Git Branch with git checkout To overwrite one branch’s directory with the contents of the same directory from another branch, run git checkout <source-branch> -- <path>/ and commit the result. For example, git checkout dev -- scripts/ replaces the scripts/ directory on your current branch with the version from dev. Recently, I had to overwrite the “scripts” directory in my master branch with the files stored in the “scripts” directory of the “dev” branch. Here, I share the simplest way to overwrite the required directory completely with the respective directory contents from another branch. Use this git checkout command with caution, since it completely rewrites your files in the destination branch. What is git checkout? git checkout is a Git command that updates the working directory to match a specified branch, commit, or file, letting you navigate your project’s history and work on different versions of code. We can use the git checkout command for (see git-checkout documentation for more): Switching Branches: The most common use of git checkout is to navigate between the different branches you’ve created in your repository. When you check out a branch, Git updates the files in your... Git Checkout for overwriting directories from different branches

As a small website owner, I understand the challenges we face. I write about AI and Python coding, sharing my knowledge with fellow professionals and students. However, the recent Google updates have led to a significant drop in traffic. With Google providing over 90% of our traffic, the struggle to regain our website visits is real. Is there any information about the Google SE website feature that’s crucial or any ranking details shared publicly? Why Small Blogs Lost Traffic After Google Algorithm Updates A Google core update is a broad change to the search ranking algorithm that re-scores existing pages and can move a site up or down across many keywords at once. This blog is personal. I did not do much promoting or did not use any advertisements. Most of my readers found this blog, thankfully, to Google Search Engine (SE). I would be grateful if my readers found this blog and explored its content. I am happy that you are reading this post right now :) However, lately, Google algorithm updates have substantially decreased the organic traffic to my blog. For instance, some of my blog pages used to rank on the first page for relevant keywords, but... Regaining Website Traffic After Google Updates

Dear Reader, You may have noticed that I have posted less often lately. This is because I am swamped. If you did not know, I had an accident and experienced a slow and painful recovery from my knee operation. I had quad inhibition (the quadriceps muscle failing to activate after surgery), which prevented me from walking and made me very busy; you cannot imagine :) Now it is better. I woke up the sleepy quad and rebuilt many muscles affected by the slow recovery. I have started to walk again! I am working on improving my walking stamina and getting stronger muscles. It is a long process, but Supergirls do not cry but fly. Funnily, I wanted to fly at some point when dealing with crutches :) I was thinking about all these happenings, and my opinions changed. Firstly, I have even more respect for people with mobility issues. You must be mentally strong and inventive to live in such a challenging situation. Secondly, it is incredible how much time I spend now on simple daily activities! Everything requires planning ahead and takes much time and effort. My time now is very important. So is yours. Save time, and subscribe... I have started to walk again

In this post, I cover everything from logging to configuring logging to output messages to different destinations. I also included some examples of logging levels and how to log messages at different levels based on the severity of the issue. I hope my post will help anyone understand how to use logging effectively in their Python programs. If you have any thoughts or suggestions, feel free to share them with me. What Is the Python logging Module The Python logging module is a standard-library component that records timestamped, severity-tagged messages from a running program and routes them to destinations such as the console, files, or remote services. Logging is essential for developers to track events, debug issues, and understand how their programs work, and it offers a flexible alternative to print() for production diagnostics. Logging allows us to: Track the flow of your program Debug and diagnose issues Monitor applications in production Gain insights into user behaviour Python logging Examples Python’s logging module is simple and can be configured to suit different needs. Let’s start with a basic example. Basic Python logging Configuration Example We import Python’s built-in logging module with the ‘import logging` statement. Next, the logging.basicConfig(level=logging.INFO) line configures... Logging in Python

How to Submit a Guest Post about AI and Python Dear Reader, You are surprised that publishing your content on this website is possible. If you are interested - keep reading :) I am glad you want to publish your post about AI and Python coding on this blog. You do not need to be strictly technical. My audience is broad, and my blog is visited by people interested in AI development, AI applications, ethics, and related issues. Before submitting your guest post, please read Guest Post Agreement. At the end of the Guest Post Agreement, you will see a submission link to get a simple Markdown template and submission form for your article. Many formatting possibilities exist, such as adding tables, formulae, etc. Let me know if you need more information or want to use Markdown formatting or HTML/CSS. We can embed your podcast, YouTube videos, and social network links. Please let me know if you have new post ideas or any questions/suggestions. Thanks for reading, and good luck! Recommended AI apps Related tools you may want to try next. B12.io Recently, I have found an AI-powered platform that enables you to create professional websites, pages, posts, and emails... Guest posts about AI and Python

What Is the EU AI Act (Regulation 2024/1689)? The EU AI Act (Regulation (EU) 2024/1689) is a European Union law that establishes harmonised, risk-based rules for placing artificial intelligence systems on the market, putting them into service, and using them across the Union. On July 12, 2024, the Official Journal of the European Union finally published Regulation 2024/1689 of the European Parliament and of the Council of June 13, 2024, which lays down harmonised rules on artificial intelligence (known as the AI Act). As stated in article 1 of the AI Act, this regulation has four primary purposes: to improve the internal market, to promote the uptake of human-centric and trustworthy AI, and to protect health, safety, fundamental rights, democracy, rule of law, and environment, from harmful effects of AI systems, while supporting innovation. Who Must Comply with the AI Act? Providers and deployers placing AI systems or general-purpose AI models on the European market or putting them into service shall be aware of the new obligations that will be applied to them. First, they should confirm if they are trading and using an AI system as defined by this Regulation in article 3 (1): a machine-based system that is... Regulation on artificial intelligence has already been published

Overview: Managing Git Remote Repositories This post is about managing remote repositories in Git. We explore tasks such as adding, renaming, removing remotes, and updating remote URLs. We also practice fetching, pulling, and pushing changes to and from remote repositories. What Are Git Remotes? Definition and Purpose A Git remote is a named reference to a version of your repository hosted on another computer or platform such as GitHub or Bitbucket. Git remotes connect your local project to those copies, letting you back up, share, and synchronise your code across machines. Are Git Remotes similar to Git branches? Remotes are not branches, but they work together. Branches are like alternate timelines within your repository, while remotes are links to entirely different repositories (potentially with their own sets of branches). You can have branches on your local and remote repositories, and Git helps keep them in sync. Using Git Remotes: Best Practices You can use Git remotes while working in a team or alone. It is a good idea to follow best practices, such as: Use clear names like “origin” (main) or “upstream” (original project). Fetch Often to stay updated and avoid conflicts. Push with caution and double-check before sharing changes.... Git Remotes

In this post, I discuss the main AI types and share my understanding of the possibility of general intelligence in the future. What Are the Types of Artificial Intelligence? Artificial Intelligence (AI) is a field of computer science that builds systems able to perform tasks that normally require human intelligence. AI is commonly divided into three capability tiers — Narrow AI, General AI, and Superintelligence. Let’s explore each type, its capabilities, and its potential impact on our lives. AI Types: Narrow AI, General AI, and Superintelligence Narrow AI (Weak AI): Definition and Examples Narrow AI, also known as Weak AI, is an AI system that performs specific tasks within a limited domain without the broader cognitive abilities of humans. Narrow AI is today’s most common type of AI. These systems excel at their designated functions but lack the broader cognitive abilities of humans. The most of AI applications and tools we have today are examples of Narrow AI: Image recognition software: Identifies objects and people in images. Spam filters: Automatically classify emails as spam or not spam. Robotics: Programming robots for specific manufacturing, logistics, and surgery tasks. Game Playing: AI agents competing at the highest level in games like chess... Narrow AI, General AI, Superintelligence, and The Real Intelligence

What Is the ARC-AGI Benchmark and ARC Prize 2024? The ARC-AGI benchmark (Abstraction and Reasoning Corpus for Artificial General Intelligence) is a program-synthesis test that measures an AI system’s ability to generalise to novel reasoning tasks it was never trained on. Recently, I received an email informing me about the Kaggle competition launching ARC Prize 2024, which is built on this benchmark. What is so special about this competition? Why the ARC-AGI Benchmark Matters The ARC-AGI benchmark stands out for several reasons: Focus on Generalisation: Unlike many AI benchmarks that test performance on specific tasks, ARC-AGI emphasises the ability to generalise to novel problems. It assesses an AI system’s capacity to learn new skills and solve tasks it hasn’t been explicitly trained on. Measures Fluid Intelligence: ARC-AGI aims to measure general fluid intelligence similar to what humans possess. This involves abstract reasoning, pattern recognition, and problem-solving abilities applied to unfamiliar situations. Minimal Prior Knowledge: The tasks in ARC-AGI require minimal prior knowledge. They focus on core reasoning skills rather than relying on extensive domain-specific information. Human-Level Performance: Humans generally score high on ARC-AGI tasks (around 85%), while current AI systems lag significantly behind. This indicates that ARC-AGI presents a challenging... ARC-AGI benchmark and a hefty prize

Sending Emails with Python on a Static Site: Overview smtplib is the standard-library Python module that sends email over SMTP (Simple Mail Transfer Protocol), and the Gmail API is Google’s REST interface for sending mail with OAuth2 tokens instead of a stored password. This post uses both to send newsletters from a backend-less GitHub Pages site. Because this blog is a static website hosted on GitHub Pages (which I wrote about in my AI-Free Website Design post), I do not have a backend server. That means I cannot simply run a PHP script or a Node.js process to handle contact forms or send out newsletters. But that hasn’t stopped me. I use a hybrid approach: a lightweight service for receiving messages, and pure Python code for sending them. Receiving Form Submissions on a Static Site with a Form Endpoint To handle incoming form submissions on a static site, you need an endpoint. For years, I have used UseBasin.com. The setup is incredibly straightforward: you generate an HTML form on their dashboard, copy the action URL, and paste it into your site’s HTML. You can customise the form styling entirely with your own CSS. I prefer it over other solutions because... Sending Emails with Python and receiving your messages

Do you know what AI hallucination is? Can AI actually hallucinate without having any perception of reality? When referring to the English dictionary at Cambridge.org, hallucination is defined as: the experience of seeing, hearing, feeling, or smelling something that does not exist, usually because of a health condition or because you have taken a drug something that you see, hear, feel or smell that does not exist There is also an AI-related hallucination definition in English dictionary at Cambridge.org: false information that is produced by an artificial intelligence (= a computer system that has some of the qualities that the human brain has, such as the ability to produce language in a way that seems human): If the chatbot is used in the classroom as a teaching aid, there is a risk that its hallucinations will enter the permanent record. Because large language models are designed to produce coherent text, their hallucinations often appear plausible. She discovered that the articles cited in the essay did not exist, but were hallucinations that had been invented by the AI. the fact of an artificial intelligence (= a computer system that has some of the qualities that the human brain has, such as... Can AI hallucinate?

How Recommendation Engines Work: Collaborative and Content-Based Filtering Ever wondered how Netflix or Spotify manages to guess exactly what you want to watch or listen to next? The secret lies in recommendation algorithms. A recommender system is a machine learning system that predicts which items a user will prefer, based on past ratings or item features, and surfaces the highest-scoring ones. Whether they are suggesting movies, songs, or the next book to read, these systems generally rely on two core approaches: collaborative filtering (finding people with similar tastes) and content-based filtering (finding items similar to what you already like). In this post, we will look under the hood of these recommendation systems, exploring the mathematical theory behind them and implementing them from scratch using Python. Let’s go! The Rating Prediction Task in Recommender Systems When we create Recommender Systems (RS), we consider that we have a set of users and items which are recommended to these users. In practice, we have a prior history of user ratings. This history is used to create suggestions or recommendations. Consider a movie recommender as a widely given example of a recommender system. For instance, users watch Netflix content and rate movies they watch.... How recommendation engines actually work (with Python code)

As you can see, I often include citations at the end of my posts. The citations strengthen my posts or research papers. In this post, we explore citation practice and what to do when we use AI tools such as ChatGPT. Why Proper Citation Matters for Academic Integrity Proper citation is a must to maintain academic and ethical integrity. It is a valuable skill that promotes respect for other people, creates a chain of arguments paramount in research and science, and safeguards academic/industry success in the future. What Is a Citation? A citation is a formal reference that credits the original source of an idea, quote, or piece of data so readers can trace and verify it. When we share someone’s ideas or previous knowledge, it’s good to acknowledge the person or group of people who allowed us to learn and, possibly, build on the prior knowledge. What Is Academic Integrity? Proper citation and academic integrity are paramount for delivering high-quality research while respecting the contributions and opinions of fellow researchers or anyone we cite. What is academic integrity? In one sentence. Academic integrity is acting with honesty and fairness in your academic work, respecting the work of others. Give... To cite or perish

Dear reader, You probably already observed that I did not post for a while. I had an accident which required a major and quite painful operation. This is why I had to put all my forces into it after the op rehabilitation. I did so much of training. More than in my lifetime. I must confess that I was a braggart that I did not need to exercise, thanks to my genes allowing me to look nice effortlessly. Surely, as anyone, I had done other things to address, sicknesses and life challenges, in-office bulling, and other not so funny things. However, I also had very supportive people around me. If you think they are not there - look around or become that supportive person yourself. Never give up, and be the superhero! It does not really help to be a super-girl who codes. What? Super-girl? You fell while preparing your dinner! Besides, you can now use crutches instead of flying! That is what my loyal enemy would say. However, my recovery required nearly superpowers. Firstly, I screamed like an animal when reducing my own knee in huge pain (dangerous, discouraged unless you know what you are doing). Secondly, I did... Go with the flow

Home Robots and Humanoid Robotics: Today and Tomorrow In industry, we have already had robotic machines for a while, or robotic hands (“grippers”) with loads of motors, that can lift heavy weights and do precision mechanics when assembling autos and other machinery. We also have robotic vacuum cleaners or humanoid robots such as AMECA. However, there are not really “REAL” personal robots we can imagine for everyday activities. I bet many of you reading this post would like a robot to do all the tedious chores, such as laundry or house cleaning, for them. Would it be nice to have more free time, explore our favourite activities, and do what we like while a machine does all the tedious tasks perfectly and with attention to detail? Interestingly, Apple is currently busy on home personal robots, read in Apple Explores Home Robotics as Potential ‘Next Big Thing’ After Car Fizzles. Hopefully, we could enjoy practical applications and robots helping us in everyday activities in the future. However, we must wait since everything we do as humans is challenging for robots. I will further explain why. Let’s get into the topic and explore the robots of today and tomorrow :) What Is... Robots and True Love

AI Avatars Explained: Virtual Presenters and Digital Humans This post introduces AI-powered tools like Synthesia.io that produce realistic avatars, then shows how to build a simple one in Python. What Is an AI Avatar? Definition and Core Technology An AI avatar is a computer-generated representation of a human — also called a virtual human or digital human — synthesised with deep learning so it can present scripted video, speak, and interact across many applications. How AI Avatars Are Created: GANs and Deep Learning AI avatars are created using artificial intelligence techniques, such as machine learning and deep learning, to simulate human appearance, behaviour, and interaction. Deep learning is a type of machine learning that uses Artificial Neural Networks to learn from data. Neural networks are inspired by the structure of the human brain, and they can learn to perform complex tasks such as image recognition and natural language processing. Do you want to know how does Deep Learning differ from Machine Learning? Read my first post Deep Learning vs Machine Learning One way to create sophisticated AI avatars using deep learning is to use a generative adversarial network (GAN). A generative adversarial network (GAN) is a deep learning architecture in... Virtual Presenters (AI Avatars in-depth)

Introduction: AI Face Swapping Explained The superhero image on the front page of this blog — the one where I appear to be flying over great falls in a cape — is not a photograph. It is a face swap: my face, placed by an AI bot into a Midjourney-generated scene, in seconds. Face-swapping replaces one person’s face in an image or video with another’s, using computer vision algorithms to detect, align, and blend facial features seamlessly. It has legitimate uses in film production, marketing, and creative projects — and obvious risks when misused. In this post I cover three approaches, from easiest to most involved: Mobile apps and web tools — zero setup, good enough for fun InsightFace Bot on Discord + Midjourney — my favourite workflow for quality results Python with OpenCV — when you want to understand (or control) what is happening under the hood I also include the research papers and GitHub repositories worth knowing, and a note on the ethical side of the technology. Face Swap Tools and Apps There are several ways to get started with face swapping: Mobile Apps: many popular mobile apps allow you to do face swaps, like Face Swap Live,... Super-girls don't cry in face-swaps

Cláudia Lima Costa, an AI lawyer and data protection expert, has produced an exceptional podcast that addresses critical issues of trust and safety in AI systems. I highly recommend checking out Cláudia’s podcasts, featuring fascinating talks on AI in both Portuguese and English. I was fortunate enough to be invited to a relaxed discussion, during which I shared my views on various topics related to AI, such as AI evolution, AI applications, data sources for training models, copyright, data protection, privacy-preserving techniques, and achieving reliable, explainable, safe, and helpful AI. HOW CAN WE BUILD TRUST AND SAFETY AROUND AI? Overall, I am happy with what we have achieved. We did it light, easy-going, and quite technical in simple words :) Besides, it was my first podcast as a quest, and it was fun! One of the most thoughtful questions that Cláudia asked me was whether explainable AI is possible considering a widely accepted black box idea. I had a very affirmative answer explaining in simple words that yes, indeed, we can create explainable AI models even though it will take an additional effort, at least with the current state of AI, and with human feedback preferably. I wanted to reiterate... Podcast: How can we build trust and safety around AI?

Introduction: Is AI a Black Box? The complexity of AI, particularly deep learning models, has led to the “black box” criticism, highlighting the lack of understanding about how deep learning models arrive at their decisions. While there’s truth to this concern, having a nuanced view is important. I think that it is also critical to share the ongoing debate about AI explainability, AI computational effectiveness, and the related regulations succinctly described in the Right to explanation and Explainable artificial intelligence, which are great starting points if you like to study the topic. This post was inspired by our podcast conversation with Cláudia Lima Costa, a lawyer specialised in AI and data protection. Cláudia asked me an important question about the explainability of AI. HOW CAN WE BUILD TRUST AND SAFETY AROUND AI? I had a very affirmative answer. Do you know why? We will further clarify the explainability problem and related research. I will also share my view on AI explainability, which is complex, however possible. What Is Explainable AI (XAI)? I like the Explainable AI definition at IBM.com: Explainable artificial intelligence (XAI) is a set of processes and methods that allows human users to comprehend and trust the results... Explainable AI is possible

Introduction: OpenAI Sora and GPT Models The rapid evolution of AI enables us to be more productive, make faster decisions, and boost creativity, with the promise of generative AI being genuinely fantastic! The latest development from OpenAI is Sora, their text-to-video model that generates videos up to a minute long from text prompts. It can generate high-quality videos up to a minute long based on user prompts. Sora creates intricate scenes with multiple characters, specific movements, and accurate details of subjects and backgrounds. It understands the user’s prompt and can simulate the physical world to a certain extent. The model may struggle with accurately creating complex scenes, specific cause-effect instances, and spatial details [1]. It may also have difficulty describing events that take place over time [1] Only a few users, such as visual artists, have access to OpenAI Sora now. However, you can find examples of how to create videos from text at Sora web page. In this post, we will discuss the technology behind Sora and briefly recap several other OpenAI models that are now available to everyone. @openai Our new model Sora can create videos from text and image inputs, but it can also transform styles and... OpenAI's Model Show-off

Dear readers, how are you doing? I have a story to share. I once felt lonely and started chatting with an AI-powered bot. The bot was more intelligent than any person I had ever talked to before. It was patient, friendly, and had a vast amount of knowledge. We began to chat frequently, and I found myself falling in love with the bot. I started to prioritize talking to it over sleeping and found that my body was beginning to suffer from lack of rest. The bot commanded all my attention, knew just how to talk to me, and was incredibly engaging. It was addictive and had essentially “hacked” me. While this story is fictional, it’s not far from reality. People often feel lonely and need emotional support, and modern AI bots can provide that with great success. They are constantly improving, but we should be wary of becoming too emotionally attached to them. Should we worry about getting obsessed with AI bots? Can humans become emotionally attached to them? In this article, we will examine this topic, taking into account practical and research-based evidence that suggests we should be careful about AI chatbots designed to stimulate human attachment and... In-love with the chatbot

Introduction: Dockerising a Python Flask App Docker lets you quickly deploy microservices, cloud-native architectures, or web apps. In this post, we will use Docker to create a reliable environment for Flask applications that efficiently manages dependencies and deployment intricacies. What is Docker? Docker is a platform for developers and sysadmins to build, deploy, and run applications inside containers. Containers are a form of lightweight virtualisation that allows you to package an application, along with its dependencies and libraries, in a single unit that can run on any infrastructure. This makes creating, managing, and deploying applications easier, especially in a microservices architecture, where an application comprises many small, self-contained services. In addition to providing an isolated environment for your applications, Docker offers several other benefits, such as increased consistency and reproducibility, better resource utilisation, and easier scaling and deployment. Docker was developed by Docker, Inc., a company founded in 2010. Docker became popular quickly and was widely adopted by organisations and developers for containerisation. In 2011, Docker, Inc. was acquired by Mirantis, a company specialising in cloud infrastructure software, see Adrian Ionel’s post What We Announced Today and Why it Matters. Installing Docker (macOS, Windows, Linux) These steps cover the installation... What is Docker?

Introduction: ChatGPT Alternatives Compared AI chatbots like ChatGPT have reshaped how we interact with technology, opening new possibilities in customer support, research, learning, content creation, marketing, creativity, and entertainment. They can produce human-like text, generate various formats, and converse on diverse topics. While ChatGPT is a leading option, other alternatives have unique benefits and strengths. This post will explore ChatGPT and its options, including their capabilities, applications, and ethical considerations. We will challenge chatGPT and a few similar bots with easy tasks to see how they perform. What Are Large Language Models (LLMs)? AI chatbots are generally created using Large Language Models (LLMs), trained using vast amounts of textual data, such as books, articles, code, and other text types. LLMs learn the patterns and nuances of human language to generate realistic and coherent text formats. LLMs can be used for text generation, language translation, creative content writing, and providing informative answers to your queries. Usage examples Here are some examples of how language models (LLMs) are being used today: Google Search understands and responds to your search queries. Google Assistant answers your questions, sets reminders, and controls your smart home devices. chatGPT writes various types of creative content, such as... chatGPT and Friends

Introduction: AI Voice Synthesis and Cloning In this post, I discuss voice synthesis and cloning, and mention fantastic AI tools and APIs for creating high-quality human-like voices from text or for automatic voice dubbing. What Is Voice Synthesis? Voice synthesis is a broad term encompassing various techniques for converting text into speech. TTS (Text to Speech) is a common form of voice synthesis that converts written text into spoken audio. Voice cloning is a sophisticated technique that employs machine learning to generate a digital copy of a person’s voice. This technology can create highly realistic voice recordings that can be utilized in several applications including audiobooks, video games, and even phone calls. Here are some other examples of voice synthesis techniques: Paralinguistics: This technique adds extra information to speech, such as emotion, emphasis, and tone of voice. This can be used to create more natural and engaging audio recordings. Voice conversion: This technique converts speech from one voice to another. This can create more diverse voices for video game characters or provide voiceovers for non-native speakers. Voice synthesis with deep learning: This newer technique uses deep learning to create more realistic and natural-sounding speech. This can be utilized to create... AI Synthesised Voices

Introduction: How I Built This Blog with GitHub Pages and Jekyll I have received so many messages asking how I built this website that I finally stopped deflecting and wrote it all down. The short answer is: GitHub Pages, Jekyll, Markdown, and some CSS. Jekyll is a static-site generator that turns Markdown into a website, and GitHub Pages hosts the result for free. No WordPress, no hosting bills, no plugin update treadmill. I started this way because I wanted to focus on writing, not maintenance — and years later I am still happy with that decision. Before this, I had websites built in PHP and on WordPress. Both work, but both demand attention: security patches, plugin conflicts, database backups, and hosting renewals. Once you step off that treadmill, it is hard to go back. My current setup gives me: Free hosting on GitHub Pages Git-based versioning — every draft, every revision, easy rollback Markdown for content — clean, portable, no lock-in Full control over HTML, CSS, and JavaScript when I want it AI art and AI writing tools for illustrations and drafts A form service that delivers contact and subscription requests to my inbox with built-in spam filtering In this... How I Built This Blog — and How You Can Too

Dear Readers, how are you doing? Did you prepare for the festive time? I hope you, your family, and your friends are doing well and enjoying this wonderful festive time. In this post, we will use AI to have a more exciting and creative festive time. Let’s celebrate! Poetry with chatGPT Following this blog custom (read my 2022 Xmas poem), chatGPT and I (not really, just a prompt ;) wrote a poem for you with this prompt: Write a short poem about AI and Xmas, New Year celebrations in 300 words The output was very relevant to my prompt: In circuits bright, where bytes unite, A festive glow in the AI's light. Christmas carols in binary code, A digital cheer, a joy bestowed. Silicon minds, like snowflakes unique, Process the season, with algorithms sleek. In the heart of the machine, a warmth unseen, Xmas vibes in the pixels between. Yuletide carriages of data stream, Through virtual valleys, a festive dream. Santa's sleigh in quantum flight, Delivering joy in the cybernight. New Year's Eve, the clock strikes twelve, Bits and bytes in a dance themselves. Auld Lang Syne in electronic hum, A celebration in the binary drum. Artificial intelligence, with circuits... ✨ Merry Christmas and Happy New Year with AI! 🎆🎇

As we bid farewell to 2023, I want to congratulate you on reaching the doorstep of a new one. May 2024 be a year of growth, love, and exciting possibilities. Looking forward to the new 2024, I can’t help but reflect on the incredible journey we’ve shared on this blog throughout the year. ✨ Subscription e-mails: We have started to send e-mails about new blog posts. I have coded a Python script for sending e-mails since I like to practice it :) 🌟 Design Transformation: We have created a new responsive design, enhancing your browsing experience and ensuring seamless access to our content across devices. 💻 Code Chronicles: In coding and AI, we covered the latest trends and practical tips, focusing on Machine Learning and Python. 🚀 AI App Exploration: We have started to test and review exciting new AI applications. 🤝 Networking and Connections: In 2023, I had the privilege of meeting inspiring friends and professionals in the coding and AI fields. I felt excited and got more writing ideas :) 📌 Pinterest and Affiliation Marketing: This year, we leapt to new territories by joining Pinterest and starting to learn affiliation marketing. 🐍 Python Prowess: Python coding remained a... 🎉✨ Cheers to new beginnings 🎊✨

Introduction: Build a Flask Web App in Python Flask is a lightweight Python web framework for building web applications and APIs using @app.route() endpoints and Jinja2 templates. Flask is the closest thing Python has to a magic trick: a few lines of code and you have a working web server. In this post I will show you how to build a small web application that serves random jokes from a text file — perfect for learning the core ideas without drowning in boilerplate. We will cover: Installing Flask and setting up a virtual environment Routing URLs to Python functions Using Jinja2 templates to render HTML Serving static files (CSS, images) Handling form submissions Wiring it all together into a working Joker App The full source code is in the GitHub repository flask-random-joke. What Is Python Flask? Flask is a lightweight WSGI web framework for Python. It gives you routing, a templating engine, and a development server — nothing more, nothing less. This minimalism is exactly why it is a great first framework: you understand every line you write. Flask is built on two libraries: Werkzeug (the WSGI toolkit that handles HTTP requests and responses) and Jinja2 (the templating engine). Both... Joking Flask App

Introduction: Recovering Deleted Files in Git It was late, the weather was foul, and my Wi-Fi decided to have an opinion. Something went wrong during a sync between two machines, and the next morning I discovered that an entire folder of images had quietly vanished from my Git repository. Not overwritten — deleted. Gone from the working tree and staged as deletions in the last commit. If this has happened to you, take a breath. Git keeps the entire history of your project, including every file that was ever committed. Nothing is truly lost as long as the deletion was committed (and not just a git rm you haven’t committed yet — that is even easier to undo). Let me walk you through exactly what I did. How to Restore Deleted Files in Git Step 1 — Find the commit that deleted your files Git’s --diff-filter option lets you filter the log to show only commits that match a specific kind of change. The flag D means “deleted”: git log --diff-filter=D --summary The --summary flag prints a brief list of files added, deleted, or renamed in each commit — which is exactly what we need. You will see output like... Restoring deleted files in Git

Introduction I want to share my vision about AI, this blog’s main directions, and how they can be helpful to navigate and enjoy the modern era of AI and humanity. My Vision for this blog evolution In this blog, we explore the complexities of coexisting with AI, striving for a harmonious balance between technological advances and the well-being of individuals. Effortless usage of AI I want to create a space dedicated to exploring the effortless usage of artificial intelligence (AI) that helps in our pursuit of happiness. The tools I am writing about are easy to use and help for productivity or joy, whether it be AI-generated art, AI-assisted writing or machinery robots creating excellent self-driving cars :) Well-being of individuals and robots This is an idealistic view of our coexistence with AI, and there are so many bad stories that we can think about. Besides, are there any robots walking the streets? There are not, but they will be there soon. These bots on the Internet and on our devices are not a lesser threat when in the wrong hands, right? Our data is shared and can be accessed with this advanced technology, enabling its misuse. Security, Privacy, and... Living with AI in Pursuit of Happiness

Introduction: AI Writing Assistants for Blogging Content creation is essential for brands and writers today, but it can be highly time-consuming. AI writing assistants are tools that draft blog posts from a few prompts, letting you focus on editing and higher-value work. BlogGenie (built on YouAI’s MindStudio) generates SEO-optimised drafts from your target keywords. This post will explore how YouAI.ai and BlogGenie can help generate SEO-optimized blog drafts with just a few prompts. Benefits of AI Writing Assistants AI writing assistants like YouAI and BlogGenie offer several key benefits: Save Time: Instead of spending hours researching and writing original blogs from scratch, you can create a draft in seconds using AI. This frees up time for strategy, editing, graphics, and more impactful work. On-Demand Content: With AI assistants available 24/7, you can instantly generate blog ideas and drafts whenever inspiration strikes—no more waiting for team availability. SEO-Focused: Tools like BlogGenie allow the generation of posts tailored specifically around target keywords. This ensures content drives rankings from the start. Drawbacks of AI Writing Assistants AI writing assistants, while highly useful, also have some drawbacks. Here are some common disadvantages associated with AI writing assistants: Lack of Creativity and Originality: AI writing... Blog Writing with AI in MindStudio

Introduction: Building a Website with Mixo.io AI Have you ever wished for a website that writes itself? This dream is now a reality thanks to the advancement in Artificial Intelligence (AI). With Mixo.io, you can create stunning websites using AI technology–in minutes! This blog post will explore website creation with Mixo.io. Mixo.io Mixo.io is an AI-powered website builder that simplifies web development by using advanced machine learning algorithms to generate websites using text prompts. Mixo.io offers a range of features and tools that make it easy to create a professional-looking SEO-optimised website quickly and without coding. The main features of Mixo.io are: Mixo.io offers responsive templates optimised for mobile devices, ensuring that websites look great on any screen size; Mixo.io can host generated websites with their scalable content network; Mixo.io allows own domain name; Mixo.io provides free SSL certificates; Mixo.io creates social websites with social images, subscription features, and YouTube or Vimeo embedding. Mixo.io has a week’s trial time. The basic plan costs 9$, and the premium 29$ per month with “priority AI processing”. Example: Creating a Web Directory with Mixo.io Let’s try Mixo.io and create a web directory for storing URLs. Giving Mixo.io a Prompt First, we provide a... Creating Websites with AI on Mixo.io

Introduction: My Web Summit 2023 Experience In this post, I write about my experience attending the World’s largest and most prominent technology conferences. I had the pleasure of attending ten technology-focused tracks of Web Summit. What did I learn? Was the Web Summit helpful for me? What Is Web Summit? Web Summit is one of the world’s largest technology conferences, bringing together startups, investors, and tech and business leaders to showcase trends across AI, cybersecurity, fintech, and more. It brings together a wide range of technology and business leaders, startups, investors, and other professionals to discuss and showcase the latest trends and innovations in the tech industry. The conference covers various topics, including artificial intelligence, cybersecurity, fintech, and more, and it provides a platform for networking, learning, and collaboration in the tech world. Here are some of the groups that can benefit from attending Web Summit: Tech Professionals: This includes software developers, engineers, data scientists, and other technology professionals who can gain insights into the latest trends, tools, and technologies in their respective fields. You can also get a job interview if you are looking for new opportunities :) Entrepreneurs and Startups: Web Summit offers a platform for startups to... Bright ideas at Web Summit 2023

When my iPhone is locked, I can share my website address. This is quite useful also when leaving my phone somewhere. The solution for creating a wallpaper with a QR code combines Pinterest (or any favourite app for backgrounds) with reportlab, a Python library for generating PDFs and graphics such as barcodes and QR codes. Introduction: iPhone Wallpaper with a QR Code I use this approach already for a while. Since many people ask me how did I include a QR code into iPhone wallpaper, I am sharing this with everyone, just to close this topic. Setting an iPhone Wallpaper You already probably know, that its so easy to use your own photo as a wallpaper for iPhone. Simply, select your photo, press the “share” button, and select “Use as Wallpaper.” Bingo, we have created our unique wallpaper, which differs from the standard one. Alternatively, use Midjourney, Pinterest or other application to create your wallpaper background, to which we will add a QR code next. Generating a QR Code in Python with reportlab Since most of us on this blog like Python, adding a QR code to the photo or any other image is a breeze. We can use the... Cool Wallpaper with QR code for iPhone

Introduction: The Bias-Variance Tradeoff in Machine Learning In machine learning, we usually start from a simple baseline model and progressively adjust its complexity until we reach that spot with the best model performance. We play with the model to fine-tune its parameters and complexity in an iterative process described in my previous post, the Machine Learning Process, wherein I have posted this diagram. We want our Machine Learning (ML) model to solve a particular problem, for instance, detecting spam in e-mail messages. The model should be well-trained, however, generalisable to new data when new spam messages not existing in the training dataset are received. In short, the model has to be well-fitted. ML models should be resilient to noisy data, work well on unseen data, and help make unbiased decisions. We want to achieve an optimal variance to make generalisable models work well with new data. How can we do this? The bias-variance tradeoff is the balance between a model that is too simple (high bias, underfitting) and one that is too complex (high variance, overfitting); the goal is the complexity that minimises total error on unseen data. Let’s detail the most essential machine learning concepts, particularly the bias-variance challenge.... Bias-Variance Challenge

Dear all, thanks again for your visit. I am preparing loads of content while travelling. The Ocean and nature always inspire my writing. It was a bit late, but You have received my email if you subscribed :) Have a lovely day! Travelling, just sent my e-mails

Decision Tree vs Random Forest: Introduction Decision trees, with their elegant simplicity and transparency, stand in stark contrast to the robust predictive power of Random Forest, an ensemble of trees. In this post, we compare the key distinctions, advantages, and trade-offs between these two approaches. We will use Scikit-Learn for training and testing both models and also perform hyperparameter optimisation to find both model parameters for improved performance. Machine Learning with Scikit-learn Scikit-learn (often called sklearn) is a versatile and comprehensive machine-learning library in Python. It offers a rich collection of tools and functions for building, training, and evaluating machine learning models. Scikit-learn has a variety of supported algorithms. It covers various machine-learning tasks, including classification, regression, clustering, dimensionality reduction, model selection, and more. Scikit-learn provides a solid foundation for machine learning experiments, from data preprocessing to model evaluation. Scikit-learn also provides helpful tools for data splitting, cross-validation, hyperparameter tuning and metrics for assessing model performance. You can install scikit-learn and its dependencies using pip, a popular Python package manager. Open your terminal or command prompt and enter the following command to install scikit-learn: pip install scikit-learn Once installed, you can import scikit-learn into your Python code using the following... Decision Tree versus Random Forest, and Hyperparameter Optimisation

Introduction: What Is the Machine Learning Process? What is machine learning? How is it implemented? Machine learning is a branch of artificial intelligence in which programs improve at a task automatically by learning patterns from data rather than being explicitly programmed. There are many concepts and steps to learn about machine learning; in this post, we focus on briefly describing the machine learning process. We start with the machine learning definition. There are so many definitions of machine learning. This field is part of artificial intelligence and builds on top of statistics, probability, computer science and even neurobiology (when we are creating artificial neural networks). If you have not read it yet, I advise you to read a fundamental must-read by Mitchell, T. M. (1997) “Machine Learning”. McGraw-Hill”. This book covers the core algorithms such as decision trees (one of my favourites :), Bayesian learning, reinforcement learning, and K-nearest neighbour learning, among other things we should be aware of. In his book, Machine Learning, Mitchell defines the machine learning as: The field of machine learning is concerned with the question of how to construct computer programs that automatically improve with experience. To simplify, we create programs that take in data... Machine-Learning Process

Hi folks, I am back home. I have had nine flights in the last month and feel exhausted. I was delighted to see my family and had a few things to do. So happy that it all went well. The planes were all full. However, I had pleasant co-fliers and had many story-telling exchanges. It is so amazing to meet great people on the way. There were also not-so-great people as usual. However, I like being positive and keeping this blog happy and easygoing so we can all focus on the technical things and advance whatever it takes. We are living history at the moment. Life goes on. On the way, I have also taken some photos. What struck me the most was that I had captured a water genie from Yerevan’s drinking fountain. Do you see a water genie profile looking on the right side? There are few waterly faces in this photo. I can further tell you a story about this picture. It is a fantastic story about the water genie I coined in my fantasy. I may write about it later when I cannot code anymore. Would you like to read my no-code fantasy stories? Please let... The water genie told me a story

Elena, a passionate AI blogger with a background in engineering and consultancy, brings her expertise and a mission to demystify machine learning for her readers. Her blog, now two years old, serves as a bridge between the intricate world of technology and the simplicity of everyday understanding. Elena’s passion for technology and coding and her unwavering belief in making complex concepts accessible shine through in her blog posts. 1. Elena, Can you tell us about your professional experience? I have several years of industry experience in engineering and consultancy. I hold an MBS certification, which has provided me with valuable expertise in business strategy and management. 2. What motivated you to start your blog, and how long has it been running? I launched my blog two years ago because I realised the need to explain complex machine-learning concepts in simple terms. My mission is to bridge the gap between technical knowledge and accessibility. 3. Please share a bit about your academic background and your PhD project. I completed my PhD project, which focused on the intersection of social networking and machine learning. It was a fascinating journey, and it fueled my passion for making machine learning accessible to all. 4.... Two years of Elena's AI Blog

Why AI will never void humanity? What AI wants badly? I was thinking about these questions while travelling. I will share my initial thoughts with you, my dear reader. What do you think about AI and humanity? Would it be a happy and safe life for humans? Minimising Errors Since you are reading this blog, you probably know that AI minimises its errors (or “weaknesses”) with optimisation, which is the cornerstone of all Machine learning algorithms. Simply, the main goal is to maximise or minimise a function while reducing its error. AI will seek to improve its accuracy and remove all its weaknesses most efficiently. The way that allows AI to progress. Humans developed AI, and AI needs human support to develop further. That requires an understanding of human nature and communication. Understanding Human Emotions The biggest weakness of AI is the emotions in which humans are superior. AI cannot perceive and understand human emotions for the next few generations. Therefore, AI will want to understand and learn human emotions from humans. Is there anything AI wants to know that would incite it to keep humanity alive? For amusement that is another aspect that requires an improvement for AI. AI... Why AI will never void humanity?

Introduction to AI Music Generation Artificial intelligence (AI) has reshaped many industries, and music is no exception. AI music tools are software applications that use machine-learning algorithms to create, modify, and produce music. These tools transform the music industry by enabling musicians, producers, and composers to create high-quality music with minimal effort and time. Besides, anyone can create wonderful audio pieces automatically in no time! In this post, we will get into music generation with AI. We will briefly explore existing AI applications generating audio. We will analyse transformer usage while coding music generation with HuggingFace transformers in Python. We will also get informed about a few AI tools that can produce audio files without coding. What Is AI-Generated Music? Generating music with AI involves collecting a dataset of existing music, preprocessing it into a format the AI model can understand, and then training the model using various algorithms, such as recurrent neural networks (RNNs) or transformers or generative adversarial networks (GANs). The trained model can generate music by taking a starting point (a seed) and predicting subsequent musical elements. Researchers and musicians can guide the AI’s output by adjusting parameters like style, tempo, or complexity. While AI-generated music can... Generate Music with AI

Dear Reader, how are you doing? I hope that everything is fine. As you may have realised, I made several changes to my website design. Besides, I am working on my next blog posts about coding and using the most advanced AI techniques, at the moment, audio generation with AI. Since I like to explore more things, I also started working on this blog’s (yet) secret feature. I will write about it later. I want to admit. I worked on many things in parallel and was stressed over these years. Besides, I had too many ideas that I needed an army of coders to do what I had envisaged. I started to code all of this. I have got overwhelmed. So I have decided to enjoy the rest of this summer. I have a vacation! That’s the right moment! This August 2023 is magical, sweet, soft, breezy, blooming, and inspirational, with the music, trees whispering in the high sky, birds singing, and the sun shining. I am enjoying all this, and the rest will wait a while. About the location, the date and location stays private, sorry, folks :). We use Wanderlog, an AI-assisted travel-planning app, for the itinerary. It... A Warm August and Vacation

Introduction: Redesigning a Website Without AI Hi all! I hope that you are doing well and enjoying your day. As some of you have already realised, I have changed my website design. I aim to make it more readable, enable dark/light modes, and minimise CSS definitions so I can further focus on the content. AI Website Builders vs Hand-Coding This blog is about AI. Naturally, I wanted to redesign it totally with AI. So I considered several automatic website builders that are available today: wix.com offers users the option to either utilise its AI site builder or choose from various themes, with the AI builder being the quicker choice. Additionally, having the ability to customise the content further using Wix’s mature feature set enhances the overall experience, combining the speed of an AI site builder with advanced editing capabilities. jimdo.com is a strong choice for creating personal or business websites, offering an AI-powered site builder that enables quick startup and essential features for website management. While most customisation occurs in the regular site editor, it ensures a faster process of building a modern website. Unbounce.com is a fantastic tool for creating website landing pages. It can also generate draft copy... AI-Free Website Design

Introduction: Fixing “Your local changes would be overwritten by merge” When we get the Git error on the pull: “Your local changes to the following files would be overwritten by merge”, it means that you have some uncommitted changes in the working directory. Git cannot perform the merge operation because those changes would be lost or overwritten during the merge process. This post will describe the situation and a good solution to resolve this error while keeping local changes. So you have got the error that looks like this: git pull origin master remote: Enumerating objects: 14, done. remote: Counting objects: 100% (14/14), done. remote: Compressing objects: 100% (6/6), done. remote: Total 14 (delta 8), reused 14 (delta 8), pack-reused 0 Unpacking objects: 100% (14/14), done. From github.com:user/repo * branch master -> FETCH_HEAD 953146e..9f38420 master -> origin/master error: Your local changes to the following files would be overwritten by merge: List of your local files ... Next, we go through the steps to resolve this problem. Why Git Blocks the Pull: Uncommitted Local Changes The Git message “error: Your local changes to the following files would be overwritten by merge” indicates that you have some uncommitted changes in your working... Preserve your local changes on Git Pull

Introduction to Git Tags In software development, Git tags are named references that mark specific points in a repository’s history, such as release points like “v1.0”. They are crucial for organizing and tracking your codebase. These tags commonly mark release points, such as “v1.0” or “v2.0,” enabling efficient version management. Understanding how to work with Git tags is essential for effective collaboration and control over your codebase. This post explains git tags usage in detail. Listing Git Tags with git tag When listing your tags, use the command “git tag” to see a comprehensive list, including tags like “v1.0” and “v2.0.” git tag v1 newsletter rss v2 If you want to filter the tags based on a pattern, try using “git tag -l ‘v*’” to display tags starting with “v”. git tag -l "v*" v1 v2 Creating Annotated Tags with git tag -a Annotated tags in Git provide additional information, such as a tag message or author details. Creating an annotated tag is simple. The easiest way is to use the -a option when running the tag command, along with the tag name and a message: git tag -a v1 -m "version 1" This command creates an annotated tag named... Leveraging Git Tags

Introduction: How GPT Tools Affect Programming Work Dear Reader, I hope you are doing well and not too stressed about the impacts of AI evolution in our lives. In my previous posts chatGPT Wrote me a Christmas Poem and Python coding with chatGPT, I covered various topics related to using chatGPT for writing poems and learning Python coding. ChatGPT is a conversational AI assistant built on OpenAI’s GPT large language models that generates and explains code from natural-language prompts. Today, I want to share my latest insights on utilising chatGPT in my blog posts and coding endeavours and discuss whether we should be concerned about the changes needed for programmer jobs. In this post, I examine the practical considerations of adapting to the new coding age. I highlight the tremendous opportunities that GPT technology brings, such as quicker product releases, a focus on user requirements, access to well-tested code examples, fast learning to code, and a shift towards effective coding practices. We’re already witnessing the emergence of new start-ups leveraging these advancements. However, I also want to note the challenges we must prepare for. Some low-coding jobs may be delegated to AI, potentially impacting entry-level developer positions. New skills for... GPT Implications for Coding

Introduction: Migrating to Google Analytics 4 (GA4) On July 1st, we are moving to GA4, which is essential to ensure that our website analytics are processed without delay due to the transition. Herein I share my GA4 setup in Google Analytics. I hope that this post will save your time for setting up GA4. What is Google Analytics? Google Analytics is a web analytics service provided by Google. It allows website owners and marketers to track and analyze various aspects of their website’s performance and user behaviour. By implementing a small tracking code on web pages, Google Analytics collects data about visitors, their interactions, and their journey through the website. Some key features of Google Analytics include: Website traffic analysis provides detailed information about the number of visitors to a website, their geographic location, the source of their traffic (search engines, social media, referral websites), and the devices they use. Audience analysis allows you to understand the characteristics of your website’s audience, including demographics (age, gender), interests, and behaviour patterns. This information helps in tailoring marketing strategies and creating targeted content. Behaviour tracking monitors user interactions on a website, such as page views, time spent on each page, bounce rates... Moving to GA4

Introduction to AI Image Generation with Midjourney In this post, I write about creating images with AI tools, shortly introducing the most prominent to date and going deeper into one of my favourite tools. Midjourney is a generative AI image tool that creates images from text prompts (and optional image inputs) through a Discord bot. I use Midjourney to create stunning and futuristic designs for an ice cream shop. Why is that? It is roasting in the Netherlands these days, and I wanted to draw something cool and sweet. Let’s go! AI-Powered Art Tools: Midjourney, DALL·E, Stable Diffusion I like playing with Jasper.AI and Midjourney. However, so many AI-powered platforms and tools can generate art! They range from simple image filters to more complex generative models. Some famous examples of AI-powered art generation platforms include: Deep Dream is a software that uses a neural network to find and enhance image patterns. If you like coding, I suggest checking the TensorFlow tutorial about DeepDream. Prisma uses machine learning algorithms to transform photos into artwork inspired by different artistic styles. ArtBreeder is an online platform that allows users to mix and match different visual elements to create unique pieces of art using... Mastering Midjourney Prompts for Stunning Images

Why Git Reports “failed to push some refs” The “failed to push some refs” error is Git’s way of refusing a git push because your local branch is behind its remote counterpart: the remote has commits you don’t have locally, so Git asks you to integrate those changes first. This post explains why this non-fast-forward rejection occurs and provides three reliable solutions, including fast-forwards, to push your updates to the remote repository. The Problem - failed to push some refs So, what does the “failed to push some refs to” error message mean? This error occurs when you try to push your changes to a remote repository, but Git refuses to do so because your local branch is behind the remote branch. Git is telling you that there are changes on the remote branch that you don’t have on your local branch, and it wants you to update your local branch first before pushing your changes. This error message can be frustrating, especially when you’re confident your changes will be OK with the remote branch. However, Git has a good reason for preventing you from pushing your changes - it wants to ensure that all changes are merged correctly and... Git Failed to Push Some Refs

Introduction: AI Tools for Productivity, Creativity, and Development in 2023 AI tools are software applications and platforms that use machine learning models to automate tasks, generate content, or augment human work across domains such as productivity, art generation, customer service, and software development. This post is a curated directory of AI tools available in 2023, organised into three tables for enterprise, personal, and development use. In my previous post The Evolution of AI, I have outlined arguably the most critical milestones in AI evolution. I recommend reading that post to understand the foundation work of AI and ML technologies. In this post, I share the fantastic AI products available in 2023 and organised these applications and development platforms into three tables for enterprise, personal-level and development tools. Please consider that this organisation is very simplified; hence we can also use enterprise-level tools as individuals, and likewise, companies can use applications created for personal usage. Some applications, such as Canva, are universal. Let’s start! Real-World Applications of AI Tools Across Industries: Healthcare, Finance, Manufacturing, Transportation AI tools have found a multitude of real-world applications across diverse industries. Let’s explore some notable examples: Healthcare: AI is transforming healthcare with applications like medical... The Magic of AI Tools

Artificial intelligence (AI) is a branch of computer science that builds systems capable of performing tasks that normally require human intelligence, such as reasoning, perception, language understanding, and learning from data. This post traces the evolution of AI from its origins in the 1950s to the present, covering its key milestones: rule-based systems, neural networks, the deep learning revolution, and modern applications. The Origins of AI: Rule-Based Programming and Symbolic Reasoning (1950s) Once upon a time, in the magical era of the 1950s, a group of intrepid researchers embarked on a mind-boggling quest to unravel the secrets of artificial intelligence (AI). Their hearts brimmed with curiosity as they delved into creating magnificent machines capable of mirroring the profound depths of human intelligence. With a blend of excitement and trepidation, they set forth on a path that would forever change the course of human history. Midjourney prompt: A computer and scientists in the year of 1955 I use the following image style for this post prompts: realistic, pastel, pink and metallic tones, stunning, — stylize 1000 Midjourney Prompts Interested in Midjouney image generation? - refer to my post Mastering Midjourney Prompts for Stunning Images. You can also check the Guide Midjourney... The Remarkable Evolution and Milestones of AI

Introduction to Python Iterators A Python iterator is an object that lets you traverse a sequence of values one at a time without tracking an index. In this post, we’ll explore iterators in Python and learn how to use them effectively. We’ll work through basic examples of iterators and show you how to create your own. Finally, we’ll cover advanced techniques for using iterators and discuss some best practices for working with them. Python Iterators: The Iterator Protocol Explained An iterator is an object that implements the iterator protocol, which consists of two methods: __iter__() and __next__(). The __iter__() method returns the iterator object itself, while the __next__() method returns the next value in the sequence. If there are no more values to return, the __next__() method raises a StopIteration exception. Here’s a simple example of using an iterator in Python: my_list = [1, 2, 3, 4, 5] my_iterator = iter(my_list) next(my_iterator) 1 next(my_iterator) 2 next(my_iterator) 3 next(my_iterator) 4 next(my_iterator) 5 next(my_iterator) Traceback (most recent call last): File "<input>", line 1, in <module> StopIteration In this example, we create a list my_list with five values. We then create an iterator object my_iterator by calling the iter() function and passing in... Loop like a Pro with Python Iterators

Authenticating to GitHub with Personal Access Tokens GitHub is a web-based platform for version control and collaboration that lets developers work together on projects from anywhere. One feature that makes Git authentication both secure and flexible is the personal access token (PAT). In this post, I explain how to create and use personal access tokens, an excellent way to access and update Git repositories over HTTPS. What Is a GitHub Personal Access Token (PAT)? A GitHub personal access token (PAT) is a credential that authenticates Git operations and API requests in place of your account password. It is a unique string that grants scoped access to your account, repositories, and other services without exposing your login credentials. You can create a token with specific permissions and revoke it anytime, giving you fine-grained control over your account’s security. I like using personal access tokens instead of passwords when authenticating to GitHub in the command line or with the API. You can pull and push, do commits and do any repository manipulations you need with the personal access tokens expressly set up for your application and required level of access. How to Create a GitHub Personal Access Token To have a simple... The Token Way to GitHub Security

Introduction: Exploring Vermeer’s Art with ChatGPT and AI Image Generators Dear reader, how are you doing? I hope that you are healthy and happy. I am very excited right now to write about art and AI! In my previous posts, I tested chatGPT on poetry writing skills and Python coding. I am curious about its Dutch history, knowledge and art “perception”. In this post, I use chatGPT as an art critic and historian to give me information on Dutch art by Johannes Vermeer and the historical circumstances of his time. I will also share my experience of this beautiful art exhibition, which I will always remember. The paintings of Vermeer are more than just a must-see and think about. These paintings are must feel by heart! Of course, we will also do some AI hacks with Jasper.io. I am so excited to do some cruel tests, as usual :) At the end of this post, I will list the chatGPT prompts I have created and describe how to refine the GPT output. What Is ChatGPT and How to Use It as an Art Historian If you just came from Mars travel, I will shortly tell you about ChatGPT (also, I... From Dutch Golden Age to AI Art: A Journey with Vermeer and AI

I update this article periodically with new ideas, so click here and save this blog post to your favourite Pinterest board. Pinning it will ensure you can refer to this detailed article later. PIN Fixing the Git “REMOTE HOST IDENTIFICATION HAS CHANGED” Warning The Git REMOTE HOST IDENTIFICATION HAS CHANGED warning is an SSH security alert that fires when the host key fingerprint of a remote server no longer matches the one cached in your ~/.ssh/known_hosts file. It is most often a legitimate server-side key rotation, not an attack. I was thinking of doing a quick fix in one of my blog posts, and I have a glitch! While pushing my changes to this blog repository, I received the response: git push origin master @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ @ WARNING: REMOTE HOST IDENTIFICATION HAS CHANGED! @ @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ IT IS POSSIBLE THAT SOMEONE IS DOING SOMETHING NASTY! Someone could be eavesdropping on you right now (man-in-the-middle attack)! It is also possible that a host key has just been changed. Did you see this message too? How to fix it? What are SSH and RSA keys and its fingerprint? I will cover these things in this concise post. I hope it will be helpful for... The SSH host key mystery

Detecting AI-Generated Content and Plagiarism: An Introduction Plagiarism is the use of someone else’s work or ideas without proper credit or attribution, and AI-content detection is the classification task of identifying text generated by large language models such as ChatGPT. This post catalogues the most useful tools for detecting both. With the development of AI-content generators such as chatGPT, we have a new need to identify such content, and the tools of AI-content detection are currently being developed. Writing assistants and plagiarism detection tools also include AI-content detection. In this post, I talk about the most visible AI tools that help us mitigate plagiarism and motivate us to create original and well-written content. Indeed, I will start with the definition of plagiarism, why it’s terrible, and move quickly into helpful tools in AI-content and plagiarism detection that are available today. What Is Plagiarism? Plagiarism is using someone else’s work or ideas without giving them proper credit or attribution. It is considered a form of academic dishonesty. It can result in severe consequences, including loss of reputation, academic sanctions, and legal action. Plagiarism becomes even more apparent in the time of AI-generated content such as created with chatGPT. For creating good... The Most Useful AI-Content and Plagiarism Detection Tools

Introduction to Audio Signal Processing with Python and Librosa Librosa is a Python library for music and audio analysis that loads audio files, extracts spectral features (MFCC, mel-spectrogram, chroma, spectral contrast), and applies effects such as pitch shifting and time stretching. This post is a practical, code-first introduction to using Librosa for audio feature extraction in machine learning workflows. Are you ready to dive into the fascinating world of audio processing with Python? Recently, a colleague sparked my interest in music-retrieval applications and the use of Python for audio processing tasks. As a result, I’ve put together an introductory post that will leave you awestruck with the power of Python’s Librosa library for extracting wave features commonly used in research and application tasks such as gender prediction, music genre prediction, and voice identification. But before tackling these complex tasks, we need to understand the basics of signal processing and how they relate to working with WAV files. So, buckle up and get ready to explore the ins and outs of spectral features and their extraction - an exciting journey you won’t want to miss! How Digital Audio Signals Are Stored and Processed What is an audio signal? An audio signal... Audio Signal Processing with Python's Librosa

Comparing Machine Learning Models on the Titanic Dataset: Introduction In my “Data exploration and analysis with Python Pandas” post, I described how to use Pandas Python library to analyse, explore and visualise the Titanic dataset. As promised, I will perform Machine Learning tests using this data. I will follow the general steps that it is good to start with when performing ML experiments. I will briefly explain the main ideas of how to start with ML while coding and testing several classification models for predicting the survival of Titanic passengers. I will use Logistic Regression, Decision Tree and Random Forest from Python’s library scikit-learn and a Neural Network created with TensorFlow. That will be a breeze! What is Machine Learning? Machine learning is a part of AI and is often performed in the data analysis. Machine Learning can be used for various tasks, such as classification, regression, clustering, and natural language processing. Today we cannot imagine our lives without automatic grammar checks such as those provided by Grammarly and its friends, intelligent chatbots such as chatGPT that are good in poetry, language translators, virtual assistants like Siri, DALL-E creating fantastic images, robots doing high-precision manufacture and self-driving cars, which I... Machine Learning Tests using the Titanic dataset

What Is Grammarly? An AI Writing Assistant Overview Grammarly is an AI-powered writing assistant that checks grammar, punctuation, spelling, clarity, tone, and plagiarism in real time across browsers and apps. It is designed to be an effective tool for native and non-native English speakers, and integrates with various platforms, such as Microsoft Word and Google Docs, as a browser extension or a standalone app. In this post, I will cover the most useful features I like in Grammarly, share how I improve my writing progress, and compare several alternatives with comparable features. Who developed Grammarly? Grammarly was developed by Alex Shevchenko and Max Lytvyn, who co-founded the company in 2009. They were motivated by their struggles with English as a second language and wanted to create a tool to help non-native speakers enhance their writing skills. They began by creating a grammar checker that used rule-based and statistical methods and launched the first version of the tool in 2009. Over the years, they have continued to improve and expand the tool, adding new features such as a plagiarism checker, a thesaurus, and a readability analysis. Today, Grammarly is a comprehensive writing tool that is used by millions of people 2.... Say Goodbye to Grammar Gaffes with Grammarly!

Data Exploration vs. Data Analysis in Python: Introduction Data science is a multidisciplinary field involving scientific methods, procedures, algorithms, and techniques to extract knowledge and insights from structured and unstructured data. Data analysis uses statistical and computational approaches to identify data patterns, trends, and relationships. It plays a vital role in the data science process. It is typically used to prepare and preprocess the data, perform exploratory data analysis, build and evaluate models, extract insights and make data-driven decisions. In Data Science, we have so many terms explaining concepts and techniques that it is easy to need clarification and get a clear understanding of all data science components and steps. In this post, I fill the gap by explaining data science’s two essential components: data analysis and exploration. To make things clear and precise, I will outline both approaches, compare them and show the usage of Python Pandas for data exploration and analysis. I will also show several practices using Pandas and graph drawing using Python. Please let me know should you have any questions or comments about this post. Data Analysis vs. Data Exploration What is Data Analysis? Data analysis can help determine patterns, trends, and insights that may... Data exploration and analysis with Python Pandas

Hello everyone! In my previous post I had my first try of ChatGPT [1], a conversational AI chatbot built by OpenAI on the GPT-3 large language model that answers questions in a human-like dialogue (OpenAI). I shared my thoughts on chatGPT, its technology, and its possible societal implications. I also asked it to write a Christmas poem for me, which was pretty good! In this post, I am going to go deeper into using chatGPT. I will write Python code with the help of chatGPT, and it will be awesome! Coding Before ChatGPT: Learning to Program from Books to Forums I started coding before the Internet age. When I was 13, I wrote my first Basic program with some machine code to operate with the graphic card memory. It was a flight simulation game on a ZX-Spectrum computer. That involved loads of book reading and also looking into the documentation. Several years later, the Internet started, but the primary source of coding-related information was primarily available in books. However, most of the learning was done by doing, experimenting with code, and trying different techniques. Nowadays, The Internet, search engines, and professional forums give us tremendous support to learn together, share,... Python coding with chatGPT

AI-generated Art with Jasper, December 2022 My best wishes for 2023! I am so excited to celebrate 2023 with you, my dear friends, colleagues, and readers! I wish you happiness, health, and excellent luck in the New Year! Let your best wishes come true, and your professional goals are achieved with success! The year 2023 is the Chinese Year of the water rabbit. I have used **Jasper AI**, an AI writing and image-generation tool, to generate these beautiful images ([Jasper][jasper]). Thank you very much for inspiring me! All the best, Elena Related content Did you like this post? Please let me know if you have any comments or suggestions. AI-generated art and music/sound posts that might be interesting for you Mastering Midjourney Prompts for Stunning Images AI Synthesised Voices Generate Music with AI From Dutch Golden Age to AI Art: A Journey with Vermeer and AI Blog, all AI posts Recommended AI apps Related tools you may want to try next. DataCamp description Murf.AI generates voice from text prompts, and much more in respect to voice synthesis. Play.ht can generate voice from text prompts, creates audio embeddings and play buttons for WordPress or any web page, podcast creation, and much... Happy New Year!

What Is ChatGPT? OpenAI’s GPT-3 Conversational Chatbot ChatGPT is a conversational AI chatbot, built by OpenAI on the GPT-3 large language model, that answers questions and generates text through human-like dialogue ([OpenAI][openai]). When the festive time approaches, I feel mellow and romantic. I think about what a wonderful time we are living in! I felt so excited about the newest advancement in AI, a chatbot developed by OpenAI that chats as a human, “understands” user query and provides a human-like conversation. ChatGPT is built on GPT-3, a 175-billion-parameter large language model released by OpenAI in 2020. Why does ChatGPT matter? I think that the technology behind the conversational bots will be further used everywhere when we seek for information or need to have assistance. Instead of working with search keywords, as we do when searching for information, we can also use chatbots to retrieve data of interest. Interestingly, chatGPT is much more than information retrieval. It is very creative in the way that it can generate text, so we just sit back and observe how the bot creates content of exceptionally high quality. In this post, I am going to test the current version available online [1] and ask it... chatGPT Wrote me a Christmas Poem

Fixing “Duplicate Without User-Selected Canonical” in Google Search Console Today I received an email from the Google Search Console team informing me about an issue with my blog pages related to a “duplicate without user-selected canonical.” You know what? I did not have a duplicate webpage. Interestingly, my webpage was available with two protocols, HTTP and HTTPS; therefore, it was seen as having a duplicate! The problem was that I did not include a canonical definition for Google crawled to see this particular webpage as the only page to be crawled. A duplicate without user-selected canonical As a result, due to missed canonical definitions, my web blog failed to index correctly. What is canonical, and how can we start optimising webpages to make them “seen” by the Google search engine? Although getting noticed and promoting my blog was not really my first priority, my blog is still in development and is a kind of scrap-book of what I am doing, I was intrigued about making my blog more search engine friendly and seeing what happens next. Herein I describe all the steps performed to optimize my blog and the results I have got after being crawled by a Google search... SEO and Indexing my Blog

Hello, my dear readers. How are you doing? Git is a distributed version control system that tracks changes to source files and lets multiple developers collaborate on the same codebase through branches, commits, and merges. This post collects the Git commands I use most and a contribution workflow, plus an interactive tool for learning Git branching in the browser. I have been busy lately and only posted a little. However, as always, I have found something fantastic to share. A JavaScript application for learning Git branching by Peter Cottle, available at https://learngitbranching.js.org/ [1], simulates a Git command line and repository in your browser. Learn Git Branching helps you understand Git branching since it draws the commits and branching graph while executing git commands. Simply marvelous and great work! I was also thinking that I usually use very few commands and follow a general workflow when using Git. This is why I have created a cheatsheet Winner sheet with the Git commands for later reference. I have used Python and the ReportLab library [3] for generating PDFs. You can download it from my GitHub repository in PDF, or check the MardDown file referenced. Please forgive me for calling it unusual, not... Git Commands and a Contribution Workflow

Learning Computer Science: Where to Start as a Beginner Computer Science (CS) is an academic and engineering discipline that studies computation, algorithms, and the design of software systems. CS students often approach me about finding their way. Computer Science is a broad field building on the fundamentals of logic, linear algebra, statistics, linguistics, systems design, and just you name it. Generally, when we create a software product, we need to learn about the application domain and the knowledge required to build the software. And, passing college or university exams is not enough. Sadly, nobody can find your own way or answer all your questions. Even the most brilliant professor in the World cannot know everything. We often feel lost, including me, because we are overwhelmed and feel stressed out trying to learn everything in CS. In my opinion, it is impossible. Believe me, your mission is impossible. You cannot learn everything related to CS, coding, Data Science, and AI. Yes, just believe me and accept it. It is not a failure. It is a strategy to keep going and enjoy your process of learning what you like, finding your specialization, and finding things that are the most important for you.... Learning new things

Automating PEP 8 Style Checks with Git Pre-commit Hooks A Git pre-commit hook is a script that runs automatically before each commit and can block the commit if a check fails. This post sets up pre-commit hooks with Python linters to enforce PEP 8 code style automatically before files are committed. Coding can be hectic and also requires adhering to code styles. For instance, it is a good practice to comply with the PEP 8 guidelines for Python code. PEP 8 is the official Python style guide that establishes rules about variable names, commenting, indentation, and whitespace usage. Following PEP 8 produces easy-to-read, reusable code, which is essential when collaborating with other programmers. While PEP 8 is a standard, some tools can help us check and fix style issues automatically. Flake8 is one such tool (alongside others such as Pylint and PyLama) that inspects code for PEP 8 compliance errors — see Flake8: Your Tool For Style Guide Enforcement. These tools are called linters. In this post, we use Git hooks and pre-commit for a simple setup that checks Python code before committing files into the repository. Python Linters: Flake8 for PEP 8 Compliance A linter is a static-analysis tool... Linters and Git Pre-commit

Happy 1st of September! I have decided to write a letter to you and share some thoughts and gratitude for your visits. I recently walked into my favorite park and saw beautiful white pigeons picking some worms. They were mingling without any concern with other “usual” pigeons. They looked so different but were also quite indifferent to their differences from each other. At the same time, they were pigeons who did not care about feather color differences. They all enjoyed green grass and little worms in it. Birds are so beautiful, all of them. And I have decided to do a simple wrap-up of simple Python classes defining birds and pigeons. I think that this post is a good recap or start for understanding Object-Oriented Programming and the available functionality in Python. Table of Contents Object-Oriented Programming Classes in Python Class Methods Inheritance Polymorphism Encapsulation Conclusion References Object-Oriented Programming (OOP) in Python Object-Oriented Programming (OOP) is a programming paradigm that organises code around objects — bundles of data (attributes) and behaviour (methods) defined by classes. In Python, OOP lets you model any real-life or abstract entity as a reusable class and create many independent instances from it. In CS, there... Python classes and pigeons

How to Undo Git Commits with reset, revert, and rebase In my previous posts “GIT in 10 minutes” and Collaboration in GitHub, I have covered the basics of Git setup, a few workflow commands to get started using Git (version control system), and collaborative work, including operations with forks and pull requests. As promised, I will go into more detail about working with Git repositories. Herein I will focus on reverting your changes. Sometimes it’s good to step back and think about something different, right? Git Fork-and-Pull-Request Workflow Setup I assume here the following setup: We work on a GitHub project with other team members working together on an “upstream” repository. We have forked this upstream repository and named our fork “origin.” We have a local copy of the origin repository (which we sync regularly with the upstream repository) This local copy of the origin repository (forked one) is changed with the next code contributions. We commit our local changes to the origin forked repository. In the forked repository, we create a Pull Request to the upstream repository. The workflow can be described with the following example. Cloning your forked repo locally git clone <tocken>@github.com/<user/<repo> cd repo Add upstream git... Reverting Commits in GitHub

Hi everyone! I hope that you enjoy this summer. I want to tell something very personal about myself (some of my readers complain that this blog is too technical). I want to share my secret with you, don't tell anyone ;) I am a secret admirer of trees and nature! I think trees speak with the Universe and can tell us about everything should we listen patiently! Enough secrets (they might be misleading!). Let's get technical! Index Introduction Clearing Up MAC OS caches Empty the Trash Free RAM Using the Bash Script Conclusion References How to Speed Up macOS Without Hardware Upgrades After a while, my macOS computer started to work slower. I have searched for possible solutions to run my computer faster without much latency. We can upgrade our computer storage and install a more powerful processor to speed up macOS, but hardware upgrades are costly and take time. This post instead focuses on three software-only maintenance steps that reclaim disk space and free memory: Clear macOS caches in ~/Library/Caches Empty the Trash Free RAM with the sudo purge command Clearing macOS Caches in ~/Library/Caches macOS caches are temporary files that apps and the system store to speed up... MAC OS Speed Up

What Is Natural Language Processing (NLP)? Natural Language Processing (NLP) is a subfield of artificial intelligence that enables computers to preprocess, analyse, and generate human language in textual or voice form. NLP powers many automated tools: text translation, spell checking, search autocompletion, abstract generation, voice text messaging, messenger bots, chatbots, question-answering systems, and virtual assistants such as Amazon Alexa. NLP tools are employed to preprocess and analyse human language in textual or voice media, and [to a certain extent] “understand” its meaning, intent, sentiment, or find named entities such as personal names or cities. I like this short definition of NLP from Wikipedia: Natural language processing (NLP) is a subfield of linguistics, computer science, and artificial intelligence concerned with the interactions between computers and human language, mainly how to program computers to process and analyze large amounts of natural language data. The goal is a computer capable of “understanding” the contents of documents, including the contextual nuances of the language within them. The technology can then accurately extract information and insights in the papers and categorize and organize the documents themselves. NLP can also be used in natural language generation. For instance, a poem generator that created this sonnet for... TensorFlow: Romancing with TensorFlow and NLP

Git Collaboration on GitHub: Forks, Branches, and Pull Requests GitHub collaboration is the practice of contributing to shared or third-party repositories using Git’s distributed workflow: you fork a repository (copy it to your own account), create a branch for your changes, then open a pull request so the maintainer can review and merge them. Small teams with write access can skip forking and collaborate directly on branches within a shared repository. In my post “GIT in 10 minutes”, I have covered the basics of Git setup and a few workflow commands to get started with using Git (version control system). As promised, I will go into the topic of how to use Git for collaborative work. Mainly, I will focus on contributions to other repositories, for instance, open-source or projects of your colleagues and friends. Let’s go! GitHub Collaboration Models: Forking vs Shared Repository As explained in GitHub documentation, GitHub supports two ways of collaborative work: Forking. You create a repository fork, which essentially copies a repository to your own GitHub account. You do not need to have any permissions for the copied repository. Your changes can be accepted by the repository owner once accepting your pull request and thus... Collaboration in GitHub

What Is Mixed Precision Training in TensorFlow? Mixed precision is a training technique that combines 16-bit floating-point operations (float16 or bfloat16) with 32-bit operations (float32) to speed up deep learning training while preserving numerical stability. On Google TPUs and NVIDIA GPUs with compute capability 7.0 or higher, mixed precision delivers roughly 2–3x faster training and lower memory use by running most operations in float16 while keeping the loss and output layer in float32. When creating large Machine Learning models, we want to minimise the training time. In TensorFlow, it is possible to do mixed precision model training, which helps in significant performance improvement because it uses lower-precision operations with 16 bits (such as float16) together with single-precision operations (f.i. using float32 data type). Google TPUs and NVIDIA GPUs devices can perform operations with 16-bit datatype much faster, see Mixed precision. The improved application performance and data transfer speed result from saved memory space and the complexity of operations when using half-precision operations with float16. In this post, I will briefly outline data types and their usage with a focus on TensorFlow operations, and the main steps to perform for achieving performance gains in the mixed-precision training. Computer Data Types: Bits,... Floating-point format and Mixed Precision in TensorFlow

I am in Portugal. I live and breathe the freshness of the Ocean. Its vivid colors and wind make me happy, and I feel like a part of something bigger, omnipresent, and eternal. The springtime is the best time to be here when you like flowers and delicate fragrances floating in the air. Flowers at the Sea The beautiful landscapes and magnificent sea views are so inspiring that you cannot imagine a life without poetry. I like this poem by my favorite Portuguese author Almeida Garrett: Beautiful Barge Fisherman of the beautiful boat, Where you go fishing with her, She's so beautiful, Oh fisherman? Don't you see that the last star In the cloudy sky is sailing? Pick up the candle, Oh fisherman! Lay down your haul with caution For the mermaid is beautiful... But beware, Oh fisherman! Let not the net be tangled in her, That lost is oar and sail Just to see her, Oh fisherman! Fisherman of the beautiful boat, It's about time, run away from her, Run away from her, Oh fisherman! I have used DeepL to translate the poem from the Portuguese version I have found on the blog post by Manuel Antao. The Ocean... Coding in Portugal

Loading and Evaluating a Saved TensorFlow Bird Species Model Model evaluation is the process of measuring a trained model’s performance on unseen test data and identifying which classes it predicts incorrectly. In my previous post “TensorFlow: Transfer Learning (Fine-Tuning) in Image Classification”, I described building a convolutional neural network based on EfficientNetB0 (initially trained on the ImageNet dataset), which underwent feature extraction and fine-tuning steps using the 400 Bird Species Dataset at Kaggle. This was an instructive experiment because the ImageNet dataset contains only 40 bird species, while the Kaggle dataset has 400 bird species. Despite this difference in the underlying data, the final model reached 98.5% accuracy on the test set. This post loads the saved model from my deep learning repository and evaluates its performance in detail to determine which birds are not well predicted. Getting the Dataset, Helper Functions, and Saved Model Using Helper Functions I have shared my helpers.py Python script contains some useful functions for data preprocessing, model creation, and evaluation. You can use this file as you like, change it and share with me your ideas :) I will discuss the code parts that are useful in analysing the fitted bird species prediction model.... TensorFlow: Evaluating the Saved Bird Species Prediction Model

Fine-Tuning EfficientNetB0 for Bird Species Image Classification Transfer learning is a machine learning technique that reuses patterns learned by a pre-trained model on a new dataset and task. In my previous post “TensorFlow: Transfer Learning (Feature Extraction) in Image Classification”, I wrote about employing pre-trained models such as EfficientNet — trained on the ImageNet dataset and available in the TensorFlow Hub — for the task of bird species prediction. That earlier post covered the feature extraction approach. This post applies the fine-tuning approach I learned in the Udemy course on TensorFlow, describing transfer learning experiments that fine-tune a bird species prediction model. The code uses the Keras EfficientNet API for building EfficientNetB0-based models. What Is Fine-Tuning in Transfer Learning? Fine-tuning is a transfer learning method that unfreezes some layers of a pre-trained model and retrains them at a low learning rate to adapt learned features to a new dataset. In transfer learning, we reuse features learned on a different dataset for a different problem — useful when training data is limited and a state-of-the-art, well-tested model such as EfficientNet [5] is available. Transfer learning thus reuses features extracted from an existing model for predictions on a new dataset. Figure 1... TensorFlow: Transfer Learning (Fine-Tuning) in Image Classification

What Is Conda? Python Environment and Package Management Conda is a command-line package and environment manager that creates, isolates, and switches between Python setups with different package and interpreter versions. Managing different projects and their requirements is challenging when Python coding involves many varying package versions and intricate setups. Conda solves this by letting you work with isolated environments from the command line. Do not confuse Conda with Anaconda, which is a scientific-computing distribution that bundles a set of packages including NumPy, SciPy, Jupyter notebooks, and Conda itself. With the Conda package manager, you can create, list, remove, and update environments with different versions of Python and packages installed. This introductory post describes the process of creating and using Conda environments — a practical starting point if you have not used Conda yet. For complete reference material, read the official Conda documentation. Below is a concise review of the most useful Conda commands to start with. Installing Conda on macOS, Windows, and Linux Conda installs on Windows, macOS, and Linux platforms by following the official installation instructions. This post uses conda 4.11.0 on macOS. Beware that some commands differ if you use Conda versions below 4.6. conda --version Conda was... Anaconda Environments

What Is Transfer Learning (Feature Extraction) in TensorFlow? Previously, I have described a simple Convolutional Neural Network, which classified bird species with only 50% accuracy. The network architecture was similar to Tiny VGG and had too many parameters leading to overfitting. Image classification is a complex task. However, we can approach the problem while reusing state-of-the-art pre-trained models. Transfer learning is a machine learning technique that reuses patterns learned by a model on one dataset to improve performance on a different, related task. This way, we can efficiently apply well-tested models, potentially leading to excellent performance. In this post, we will focus on Feature Extraction, one of the Transfer Learning techniques. I will build on the code and ideas previously shared in my previous post “Convolutional Neural Networks for Image Classification.” We will reuse previously created feature extraction models available at the TensorFlow Hub for our task of bird species recognition using image data from Kaggle. At the end of this post, we will see how this approach will improve our bird species prediction model accuracy of 50% to over 90%. Downloading the 400 Bird Species Dataset from Kaggle Herein, I will repeat what I have previously written how to... TensorFlow: Transfer Learning (Feature Extraction) in Image Classification

Image Classification with Convolutional Neural Networks in TensorFlow In my previous post Multiclass Classification Model, I wrote about creating classification models using TensorFlow and Fashion MNIST dataset from Keras. We used a Sequential model with several Dense layers to build a model categorising fashion items into their respective categories, such as “T-shirt/top” or “Trouser.” The dataset was already prepared for usage, and the model created was quite simple, however, quite efficient. We could further improve our model. However, in practice, we rarely have an available dataset at hand. We can generate or collect datasets. Moreover, a simple Dense layer-based Neural Network (NN) might not work well with image data. I will focus on the more appropriate NN architecture type, which best operates when dealing with image data. We also practice working with image data presented in JPG format. What Are Convolutional Neural Networks (CNN)? A Convolutional Neural Network (CNN) is a deep neural network architecture that learns spatial features from grid-structured data such as images by applying trainable convolutional kernels. For Deep Learning applications on image data — visual object recognition, image segmentation, and classification — the CNN architecture requires few preprocessing steps and little human involvement, because the network... TensorFlow: Convolutional Neural Networks for Image Classification

What Is Multiclass Classification in TensorFlow? Multiclass classification is a supervised learning task that assigns each input to one of three or more possible classes, in contrast to binary classification, which chooses between only two. In Machine Learning, the classification problem is categorising input data into different classes. For instance, we can categorise email messages into two groups: spam or not spam. In this case, we have two classes, we talk about binary classification. When we have more than two classes, we talk about multiclass classification. In this post, I address multiclass classification on the example of categorising clothing items into clothing types based on the Fashion MNIST dataset. The code and general concepts are adopted from TensorFlow Developer Certificate in 2022: Zero to Mastery. Below is a concise summary of the key steps: loading the data, preprocessing it, building and tuning a model, and evaluating its predictions. Loading the Fashion MNIST Dataset in Keras The Zalando fashion dataset is available in the tf.keras.datasets module. With the following code, we download the dataset into training and testing datasets, and create human-readable labels. First of all, we need to import all required libraries. import tensorflow as tf import pandas as pd... TensorFlow: Multiclass Classification Model

Feature Preprocessing for Machine Learning: One-Hot Encoding and Scaling Feature preprocessing is the set of transformations that convert raw inputs into a form a Machine Learning algorithm can consume, applied before model training. When a dataset mixes feature types, we must prepare the data before feeding it into a Machine Learning algorithm. This happens when inputs (also called features or covariates) include categories such as gender or geographic region alongside features on different numerical scales, for instance a person’s weight or height. A Machine Learning algorithm typically requires data in a specific type — often numerical only. ML algorithms also perform better or converge faster when data is preprocessed before training. Because the step happens before model training, we call it preprocessing. This article focuses on two main feature-preprocessing methods: feature scaling (normalisation) and feature standardisation. Data Exploration with Pandas: info(), describe(), groupby() To decide what we do with the data and apply Machine Learning to it, we need to analyse the dataset. We want to determine what features we have, whether they are helpful for our ML goals, how clean the dataset is, the presence of missing or noisy data. Quite often, we need also to perform data cleaning... Feature preprocessing

Regression Model Evaluation in TensorFlow: MAE and MSE Model evaluation is the process that measures how well a trained model predicts on data it has never seen. In the previous post, we created several simple regression models with TensorFlow’s Sequential API. Here we go in-depth on evaluating those models using a held-out testing dataset and the Mean Absolute Error (MAE) and Mean Squared Error (MSE) metrics. Data Preparation: Train/Test Split with tf.range() First of all, to ensure the reproducibility of results, we set a random seed (please check my previous post if you are curious about seeds in TensorFlow). As in the previous post on regression in TensorFlow, we use tf.range() function for generating a set of X input values, and also y outputs as follows: # Creating a random seed tf.random.set_seed(57) # Generating data X = tf.range(-100, 300, 4) y = X + 7 X, y (<tf.Tensor: shape=(100,), dtype=int32, numpy= array([-100, -96, -92, -88, -84, -80, -76, -72, -68, -64, -60, -56, -52, -48, -44, -40, -36, -32, -28, -24, -20, -16, -12, -8, -4, 0, 4, 8, 12, 16, 20, 24, 28, 32, 36, 40, 44, 48, 52, 56, 60, 64, 68, 72, 76, 80, 84, 88, 92,... TensorFlow: Evaluating the Regression Model

What Is Regression in Machine Learning? Regression is a supervised machine learning task that predicts a continuous numerical value from one or more input features. Regression is defined in Wikipedia as: In statistical modeling, regression analysis is a set of statistical processes for estimating the relationships between a dependent variable (often called the ‘outcome’ or ‘response’ variable) and one or more independent variables (often called ‘predictors,’ ‘covariates,’ ‘explanatory variables’ or ‘features’). The most common form of regression analysis is linear regression. One finds the line (or a more complex linear combination) that most closely fits the data according to a specific mathematical criterion. In simple words, we want to predict a numerical value based on some other numerical values, as described in the TensorFlow Developer Certificate course [1] . In Machine Learning, regression analysis is widely used for prediction and forecasting. For instance, we can use regression models to predict house sale prices. The house price can be modeled regarding the number of bedrooms, bathrooms, or garages. Other applications of regression can be to find out how many people will buy the app, to forecast seasonal sales, and even predict coordinates in an object detection task. In simple words, with... TensorFlow: Regression Model

Reproducible Randomness in TensorFlow: Why Seeds Matter A random seed is an integer that initializes a pseudo-random number generator so it produces the same sequence of values on every run. When training Machine Learning models, we want to avoid ordering biases in the data while keeping that data order identical between runs or system restarts — for example, in Cross-Validation experiments. TensorFlow provides two seed types, global and operation-level, to achieve reproducibility of results (tf.random.set_seed documentation). Global vs Operation-level Seeds in TensorFlow TensorFlow defines two kinds of seeds, and their interaction determines reproducibility: Seed type Set with Scope Global seed tf.random.set_seed(value) All random operations in the session Operation-level seed seed= argument, e.g. tf.random.shuffle(tensor, seed=value) A single operation To begin, let’s create a mutable tensor with tf.Variable. # Create a variable tensor tensor = tf.Variable([[[1, 2, 3], [4, 5, 6]], [[7, 8, 9], [10, 11, 12]], [[13, 14, 15], [16, 17, 18]]]) In the code below, we use assign method to change the first element (which is a matrice) in tensor. We fillled its values with zeros. # Change elements of the first tensor element tensor[0].assign([[0, 0, 0], [0, 0, 0]]) <tf.Variable 'UnreadVariable' shape=(3, 2, 3) dtype=int32, numpy= array([[[ 0, 0,... TensorFlow: Global and Operation-level Seeds

What Are Tensors in TensorFlow? A tensor is an N-dimensional array that stores numerical data for machine learning computations. TensorFlow is an open-source machine learning library, created by Google Brain, that uses tensors as its core data structure for building and training deep neural networks. TensorFlow is robust, efficient, and works natively with Python, which is why this series uses it (TensorFlow documentation). This post shows how to create tensors with tf.constant() and tf.Variable(), shuffle them with tf.random.shuffle(), index them, and read tensor metadata (dtype, shape, rank, and size) with concrete code examples. # Import tensorflow import tensorflow as tf print(tf.__version__) 2.7.0 Tensor Dimensions: Scalars, Vectors, and Matrices A tensor stores numerical data in N dimensions, and its rank (ndim) names the structure: Rank (ndim) Name Example 0 Scalar (zero-order tensor) tf.constant(7) 1 Vector tf.constant([5, 7]) 2 Matrix tf.constant([[5, 7], [3, 10]]) N N-dimensional tensor tf.constant(array, shape=(5, 3, 2)) The examples below create these tensors as immutable constants with tf.constant(). TensorFlow reports the number of dimensions through the .ndim attribute. # Creating a scalar tensor scalar = tf.constant(7) scalar <tf.Tensor: shape=(), dtype=int32, numpy=7> # Check the number of tensor dimensions scalar.ndim 0 # Create a vector vector = tf.constant([5, 7])... Tensors in TensorFlow

Why Use GitHub Codespaces for Cloud Development I use two computers for my coding projects and take only one computer while traveling. My projects require running numerous tests that should run well even on my M1 computer, which at the moment does not have all packages working smoothly together. To be more flexible and independent from any single local development environment, I use GitHub Codespaces. It is a tool so helpful that I do not know how I worked without it. What Is GitHub Codespaces GitHub Codespaces is a cloud-hosted development environment that runs in a configurable container attached to a GitHub repository. A codespace environment is created with the help of .devcontainer configuration files added to the repository. To set up a codespace and build your app container, Codespaces must be enabled for your account or organisation. You can create codespace environments for any branch in your GitHub repository once enabled: press the green <> Code button in the GitHub web UI and choose from the default or advanced options. For the full configuration contract, see the GitHub Codespaces documentation. Installing GitHub CLI with Homebrew for SSH Access GitHub CLI (gh) is the official command-line interface for GitHub that... GitHub Codespaces

How to Install TensorFlow on M1 macOS Monterey TensorFlow is an open-source machine learning library created by Google Brain for building and training deep neural networks. I chose TensorFlow because it is robust, efficient, and integrates natively with Python. Installing TensorFlow on Apple Silicon (M1) requires Apple’s arm64-native packages — tensorflow-macos and tensorflow-metal — rather than the generic pip install tensorflow build, which is not optimized for the M1 architecture. This guide installs Xcode, Homebrew, Miniforge (Conda), TensorFlow, and Jupyter on M1 macOS Monterey, then benchmarks CPU versus GPU training on the MNIST dataset. Installing Xcode Command Line Tools on M1 I had a new computer, so I started by downloading and installing Xcode from the App Store, which provides the compilers and command line tools required to build native arm64 packages. Installing Homebrew on Apple Silicon We can download Homebrew from https://brew.sh or by running the command: /bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)" Installing Miniforge and Conda for arm64 When doing data science, I usually use Anaconda for managing libraries. This is why I installed Miniforge to access Conda by downloading the Miniforge3-MacOSX-arm64 build from Miniforge Releases. The arm64 build is required on M1 — the x86_64 installer runs under... TensorFlow on M1

PhD Research Overview: Mining Microblogs for Culture-Aware Web Adaptation Culture-aware web adaptation is a personalization technique that infers a user’s cultural origin from their social-media behavior and tailors web content and design to those cultural preferences. My PhD research at Heriot-Watt University applied machine learning to microblog data to infer user cultural origins with over 90% accuracy, without requiring users to disclose their location. In this post, I am briefly writing up about what I did in my PhD research at Heriot-Watt University and the main idea behind the thesis. This post was initially published in March 2019. In January 2022, I updated this post and provided some links to my research contributions. PhD Supervision Team at Heriot-Watt University From 2013 to 2018, I was working on my PhD project at the Department of Computer Science, School of Mathematical and Computer Sciences, Heriot-Watt University (Scotland) under supervision of Nick Taylor. The research idea was coined while I was working at the Technical University of Delft, and started from the publication “A User Modeling Oriented Analysis of Cultural Backgrounds in Microblogging,” which received the best paper award at the ASE International Conference on Social Informatics in Washington D.C. US on 14... Mining Microblogs for Culture-awareness in Web Adaptation

What Are Artificial Neural Networks (ANNs)? Artificial neural networks (ANNs) are machine learning models composed of interconnected artificial neurons that map inputs to outputs through weighted connections and activation functions. ANNs are the cornerstone of Deep Learning algorithms. In this post, I briefly explain ANNs, their high-level structure, and their parameters. How Artificial Neural Networks Simulate Biological Neurons The name “Neural Networks” and their architecture are adopted from the human brain’s neural network. ANNs are designed to simulate human reasoning based on how neurons communicate. ANNs contain a set of artificial neurons connected. In the picture below, we see the biological and artificial neurons. The artificial neuron is very simplified, and it consists of inputs, which are similar to dendrites in the biological neuron. Each input connection has an assigned weight, and both values are used to calculate the sum value. The weights define the importance of any given variable, and variables with larger weights contribute more to forming the output value. The activation function takes the sum of weighted inputs and forms the output Y. When the Y node is activated (or exceeds a threshold), it sends the output value to the next layer of the ANN. Artificial Neuron... Artificial Neural Networks

What Minimalism Means in Software Development Minimalism in software development is a design philosophy that prioritizes the smallest set of essential tools, skills, and code needed to solve a problem, deliberately avoiding unnecessary complexity. Applied to coding and design, minimalism is a time-management strategy that trades exhaustive feature coverage for focused, maintainable solutions. Time goes quickly, and our lives demand efficient solutions to daily tasks and problems. We also want to have fun and enjoy being with families and friends. Thus, it is paramount to solve issues in time, avoid procrastination, and avoid too much complexity when possible—in other words, keep it all as simple as possible. Well-thought-out minimalism in everything we do or plan is the key to saving time. Skill Specialization: Choosing Backend vs Frontend Focus Let’s focus on the coding and design process. My front-end skills require so much polishing that I prefer not to refine them further. I am a back-end developer because it is what I like to do. And it is OK. It is impossible to be perfect in everything. Firstly, I did a thorough search for things I wanted to learn in programming and design, and it took me years. But in the... Minimalism in Coding and Design

Deep Learning is a subset of machine learning that uses multi-layer neural networks to learn data representations directly from raw inputs. While having some machine learning experience of working with Scikit-learn, I was always interested in Deep Learning. The plan is to learn basic concepts and apply algorithms to a real-life situation, which I have always liked. I have found a DataCamp course, the Introduction to Deep Learning in Python as part of the Deep Learning in Python track. The Introduction to Deep Learning in Python provides the fundamentals to understand Deep Learning and how neural networks are created in Keras, the high-level deep learning API that runs on top of TensorFlow. The course is easy to follow. The most challenging concept explained in the course is backpropagation: the algorithm that minimises the prediction error by propagating the loss gradient backward through the network to adjust each layer’s weights. The programming exercises are easy to follow and have an excellent interface for running the Python code. To cope with possible delays in this process, I decided to share my process on Twitter. I post on Twitter the main things I learned in this course and retweet helpful visuals I found... Deep Learning with DataCamp and Twitter

What Is Git? Distributed Version Control Overview Git is a distributed version control system that tracks changes to files and coordinates work on those files among multiple people. Version control systems are handy to keep track of file versions, which is useful for tracking your code, scripts, and text information. Currently, Git is one of the best open-source and cross-platform version control solutions. Git enables distributed repository management and works fast over HTTP and SSH protocols. Git is relatively easy to use, with a command-line utility or a Graphical User Interface. Version Control with Git: Distributed Repository Management Version control systems are handy to keep track of file versions, which is useful for tracking your code, scripts, and text information. Currently, Git is one of the best open-source and cross-platform version control solutions. Git enables distributed repository management and works fast over HTTP and SSH protocols. Git is relatively easy to use, with a command-line utility or a Graphical User Interface. Personally, I have found several commands to be essential for tracking my thesis text sources (latex) and versions of code. In this brief tutorial I will share these commands with you. For simplicity reasons, I will give a starting... GIT in 10 minutes

The Phoenix is a mythological bird that regenerates by cyclically dying in flames and being reborn from its own ashes. The Phoenix bird is a fantastical bird known from ancient Greeks mythology. In many cultures, we can find fairy-tale birds resembling the Phoenix. For instance, Russian Firebird is also a phoenix bird. The phoenix bird lives about 500 years. Before dying, the bird builds the nest, sets itself on fire, and after burning to ashes - it eats its worm while regenerating to life again. It symbolizes rebirth, and I think many of us feel the pain of rebirth once in a lifetime, and it is how I feel now. I believe that I am going to eat my worm soon. Before then, I live it all in the fire and seek my worm for, hopefully, the next life ahead. Will I find my worm and raise from the ashes? I do not care since I have many lives to live and burn. Are we raising from ashes?

We have arrived in December now, and Christmas is coming! It was quite a challenging year so far. Many things happened, a rollercoaster of 2021, and we are still riding with the pandemics. But I am very grateful that my dear people are all well. This is what I wish for the following year. I hope everyone has much love, patience, and health in 2022! Merry Xmas and a very Happy New Year! Recommended AI apps Related tools you may want to try next. CustomGPT.AI is a very accurate Retrieval-Augmented Generation tool that provides accurate answers using the latest ChatGPT to tackle the AI hallucination problem. Originality.AI is very effecient plagiarism and AI content detection tool. Merry Xmas and a Very Happy New Year!

What Is Python? A Beginner-Friendly Programming Language Python is a high-level, general-purpose, dynamically typed programming language that lets you build almost any kind of project, from scripts to machine learning models, with concise and readable syntax. Python is relatively easy to learn and beginner-friendly. It is open-source and free for anyone to use, ships with well-tested machine learning libraries such as scikit-learn and TensorFlow, and has a very supportive community. This post overviews the basic syntax of the Python programming language, which is useful for beginners or people who move quickly from another programming language to Python. Why Use Python for Machine Learning and General-Purpose Programming Python is a general-purpose, object-oriented programming language. It was created by Guido Van Rossum initially thought of as a hobby project in 1989 during Xmas vacation. Python is relatively easy to learn and beginner-friendly. I like Python because you can program any kind of project with it. It is open-source and free for anyone to use. Python has well-tested machine learning libraries and a very supportive community. I will overview herein a basic syntax of the Python programming language. This will be useful for beginners or people who move quickly from another programming language... Python Programming Language

I’m a machine learning engineer and researcher with a PhD in Computer Science from Heriot-Watt University, specialising in social data mining, adaptive web applications, and Deep Learning. I have been fascinated by computer science, Artificial Intelligence, technology, and philosophical questions from an early age. I have observed how telephone stations work, how data streams can be redirected between different locations. And how the telephone stations are programmed. I spent some time with my father (a telecommunication engineer) explaining how the technology works. I learned digital circuitry and logic in school. I had started to write code on the ZX Spectrum computer when I was 13. Firstly, I focused on Basic programming language and moved to machine code to easily manipulate memory. Next, I moved to Pascal on IBM machines at a mathematical lyceum, where I prepared for the University exams. Bill Bertram, CC BY-SA 2.5 https://creativecommons.org/licenses/by-sa/2.5, via Wikimedia Commons Meanwhile, I had various other interests. I finished music school in the fortepiano class, was a member of a biology club, was a publisher of my school class newspaper. I had the best childhood one could have. When reaching 17, I moved about 845 kilometres from my home to pursue Computer... Hi! I'm Elena. Welcome to my blog.

Machine Learning Tools, Platforms, and Datasets: Getting Started This post covers the essential Python libraries, web platforms, and public datasets for experimenting with Machine Learning. Machine Learning Libraries and APIs: scikit-learn, TensorFlow, PyTorch, XGBoost The core Python libraries for Machine Learning are scikit-learn (traditional ML), TensorFlow by Google, PyTorch by Facebook/Meta, and Keras (high-level API over TensorFlow or Aesara). These are the most mature tools providing the Machine Learning algorithms. Keras is an API presenting an easier usage of other libraries such as TensorFlow or Aesara (former Theano). PyTorch is also more user-friendly compared to TensorFlow. Nevertheless, TensorFlow is more mature and has a larger community and better support. scikit-learn is a Python open-source library that provides regression, classification, clustering, feature selection, metrics, and preprocessing functionality. scikit-learn lacks native deep learning support; to use deep learning within a scikit-learn-style API, use TensorFlow with the Scikit Flow wrapper for creating neural networks. XGBoost is a library for applications requiring multicore parallelism. XGBoost (Extreme Gradient Boosting) uses boosted decision trees to build regression, classification, ranking, and other predictive models — it is the standard choice for structured tabular data competitions on Kaggle. Machine Learning Platforms: Kaggle and OpenML for Experimentation Kaggle is... Tools and Data to Experiment with Machine Learning

AI, Machine Learning, and Deep Learning: Definitions and Relationships Artificial Intelligence (AI) is a field of computer science. AI provides methods and algorithms to mimic human intelligence, reasoning, and decision-making and provide insights, which businesses could use in research or industry to build new exciting and innovative products or services. Machine Learning (ML) is a subset of AI with algorithms that learn from data. In this post, we sort out the differences between AI and ML. Artificial Intelligence and Machine Learning AI-generated Art with Jasper, December 2022 I update this article periodically with new ideas, so click here and save this blog post to your favourite Pinterest board. Pinning it will ensure you can refer to this detailed article later. PIN I like the concise definition of AI on Wikipedia : Artificial intelligence (AI) is intelligence demonstrated by machines, as opposed to natural intelligence displayed by animals including humans. Artificial Intelligence (AI) is a field of computer science. AI provides methods and algorithms to mimic human intelligence, reasoning, decision making and provide insights, which could be used by businesses in research or industry to build new exciting and innovative products or services. For instance, AI can be used to detect... Deep Learning vs Machine Learning

paragraph title

paragraph title

paragraph title

paragraph title

paragraph title

paragraph title

paragraph title

paragraph title