AI Terms Explained

AI headlines are full of words nobody defines for you: parameters, tokens, quantization, open-weight, context window. This series translates them, one everyday analogy at a time, starting from absolute basics and working up to the vocabulary you’ll find in a real model release.

What You’ll Learn

• Part 1: The Building Blocks — model, parameters, training, inference, tokens, and prompts
• Part 2: How AI Models Learn — training data, fine-tuning, overfitting, and RLHF
• Part 3: Key Machine Learning Concepts — the maths and algorithms under the hood, for readers who want to go a level deeper
• Part 4: Key Concepts in AI — neural networks, attention, transformers, and modern LLMs
• Part 5: Shrinking Giant Models — quantization, GGUF, MXFP4, RMSNorm, and precision
• Part 6: Reading a Model Release — open-weight, SOTA, native vision, and context windows, decoded from a real release
• Part 7: The AI Terms Cheat Sheet — every term from the series in one quick-reference glossary

Series Progress

All Posts in This Series

Part 1: AI Terms 101: The Building Blocks Coming Soon What is a 'model', a 'parameter', or a 'token', really? Part one of a new series translates the words you keep seeing in AI headlines into plain English, one everyday analogy at a time. This post is currently being written and will be published soon.
Part 2: How AI Models Learn: Training Data, Fine-Tuning & Why Models Get Things Wrong Coming Soon Training data, epochs, overfitting, fine-tuning, RLHF: the words behind how a model goes from blank slate to (hopefully) useful assistant, explained without the maths. This post is currently being written and will be published soon.
Part 3: Key machine learning concepts Coming Soon A practical machine-learning concepts guide with math intuition and implementation-focused checks to avoid common modeling errors. This post is currently being written and will be published soon.
Part 4: Key Concepts in AI: A Comprehensive Guide to Artificial Intelligence Fundamentals Coming Soon A practical AI fundamentals guide connecting key concepts to implementation choices and real engineering tradeoffs. This post is currently being written and will be published soon.
Part 5: Shrinking Giant Models: Quantization, GGUF & Precision Explained Coming Soon Quantization, GGUF, MXFP4, RMSNorm, full-precision inference, lossless — the vocabulary of making trillion-parameter models small enough to actually run, explained with everyday comparisons. This post is currently being written and will be published soon.
Part 6: Reading a Model Release: Decoding Release-Day Jargon Coming Soon Open-weight vs closed, SOTA, activated parameters, native vision, 1M-token context windows — using a real release (Kimi K3) to translate an entire model announcement into plain English. This post is currently being written and will be published soon.
Part 7: The AI Terms Cheat Sheet: Your Quick-Reference Glossary Coming Soon Every term from this series in one scannable reference: model, parameters, tokens, quantization, open-weight, context window and more, in plain English, in one place. This post is currently being written and will be published soon.

Getting Started

This series is coming soon. The first post introduces the foundations.

Who Is This Series For?

This series is for anyone who reads AI news and quietly wonders what half the words mean — no machine learning background required for the earlier parts. Later parts get more technical, but I flag that clearly as we go, so you can stop wherever the depth suits you.

Series Philosophy

• Analogies over algebra. Every hard idea gets a physical comparison before anything else.
• Beginner to advanced, honestly labelled. Parts 1 and 2 assume nothing. Parts 5 and 6 assume you’ve read the earlier ones.
• Grounded in real releases. Where possible, terms are illustrated with an actual model announcement, not a made-up example.