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.

Most of the 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 and Go.
Recommendation engines: Suggest products or content based on user preferences.
Self-driving cars: Navigate roads and make driving decisions.
Financial trading: Predicting stock markets and making automated trading.
Large Language Models: Process and generate human-like text in response to a wide range of prompts and questions, summarise and translate human text, converse with humans and answer questions based on the knowledge base.

In-love with the chatbot discusses AI companions, chatbots and a danger zone in LLMs applications.

Please note that although Large Language Models (LLMs) such as ChatGPT and Google Gemini can perform impressive feats in language generation and comprehension, they are still considered Narrow AI. This means that they are specialized in specific tasks and lack the broad cognitive abilities of humans, such as general reasoning, common sense, and consciousness. LLMs operate based on patterns they have learned from the data they were trained on, and they struggle to grasp the actual meaning behind the words they process.

LLMs represent a significant advancement in AI, but they are still far from achieving the level of General AI or Strong AI, which would possess human-like intelligence across a wide range of domains.

General AI (Strong AI): Definition and Capabilities

General AI, or Strong AI or Artificial General Intelligence (AGI), is a hypothetical type of AI with human-level intelligence across various domains. It can understand, learn, and apply knowledge similarly to humans. OpenAI’s own charter defines AGI as “highly autonomous systems that outperform humans at most economically valuable work” — a deliberately economic bar rather than a philosophical one. Google DeepMind, for its part, proposes levels of AGI based on performance and generality, because nobody in the field actually agrees on a single test for “human-level.” While General AI remains a theoretical concept, its development is a significant goal in AI research.

Potential Capabilities of General AI:

Problem-solving: Tackle complex problems and find innovative solutions.
Natural language understanding: Comprehend and respond to human language meaningfully.
Creativity: Generate original ideas and artistic expressions.
Reasoning: Draw logical conclusions and make informed decisions.
Learning from own and other experiences (for instance, from videos or stories): Learns from experiences and builds strategies in similar situations.

In my post about the ARC-AGI benchmark, I share my honest opinion about the possibility of AGI soon. However, I write primarily about the incredible Kaggle competition, which you can join to study AGI further while coding and having fun, potentially getting a good prize for your efforts.

Update: Six months after I first wrote this, OpenAI’s o3 model jumped to 87.5% on that same ARC-AGI-1 benchmark (up from o1’s 32%), which pushed the benchmark’s creators to release a much harder ARC-AGI-2 test in 2025. A good reminder that these goalposts move fast, and a single benchmark score is not the same thing as AGI.

While a true AGI has yet to be developed, some argue that specific AI systems exhibit characteristics associated with AGI, albeit in limited domains. These include:

Deep Reinforcement Learning: AI agents learn complex strategies through trial and error in simulated environments, like OpenAI Five’s mastery of Dota 2.
Generative AI: Systems like ChatGPT generate realistic human-quality text but still need proper understanding and reasoning.
Multimodal Learning: AI systems process and understand information from various sources, such as text, images, and audio, progressing in areas like question answering.

Superintelligence: Definition and Implications

Superintelligence is a hypothetical type of AI that surpasses human intelligence in virtually every aspect. Philosopher Nick Bostrom, whose early writing on the topic helped shape the field, describes it as an intellect that greatly exceeds the best human performance in practically every domain of interest. It would possess cognitive abilities far beyond our own, potentially leading to transformative changes in our world. Nobody knows if or when superintelligence will exist, but its potential impact fuels no end of debate.

Potential Implications of Superintelligence:

Scientific breakthroughs: Accelerate research and development in various fields.
Economic disruption: Automate jobs and transform industries.
Existential risks: Pose potential threats to humanity if not controlled.

In my previous post Living with AI in Pursuit of Happiness, I share my opinion on the implications of evolved superintelligence. Indeed, all these risks and impacts will be challenging. Still, they also enable benefits for humanity, such as living that fosters human well-being, curing dangerous diseases, focusing on creativity and happiness, and technological advancement. Just think about the evolution of manufacturing, which was initially challenging, but it has brought some goodness to our lives.

The Internet and all the devices you use were only possible with the Industrial Revolution and the efforts of previous generations. AI will provide a toolset to improve our lives further. Should we worry about its consequences? We should work on many aspects and refrain from delegating our responsibility.

You can also check my related post on robotics and the current research issues explored in Robots and True Love.

A Few Remarks on Superintelligence and AGI

How AGI Could Lead to Superintelligence

In short, AGI could understand and learn any intellectual task a human can, not just specific ones. It could adapt to new situations, solve problems independently, and even have its own thoughts and feelings. This level of AI is still theoretical and far from being achieved, although research is steadily progressing.

Moreover, if AGI exists, it could surpass human intelligence in all aspects, becoming Artificial Superintelligence. Superintelligence carries potential risks, so careful consideration and ethical development are crucial if we ever reach this stage.

How Superintelligence Goes Beyond Human Abilities

Eventually, superintelligence will go beyond human reasoning abilities. It would excel in more than one area, like playing chess or recognizing images. It would possess generalized intelligence, potentially capable of understanding, learning, and reasoning at a level far exceeding any human.

This has unpredictable potential and profound implications. It could bring unimaginable benefits, like solving complex problems or accelerating scientific progress. However, it also raises serious concerns about possible risks and unintended consequences.

Many questions still need to be answered. Could we control such an AI? Would it have goals and desires similar to humans? Would it see us as allies or threats?

We need to prepare for advanced AI, build it safely, and use it to improve our lives — while weighing the ethical issues, the risks, and the bigger philosophical questions about humanity’s future.

Why Do We Need Superintelligence?

It is simple. We may need superintelligence for human survival. There is a non-zero possibility that Earth could become uninhabitable, or that it will be destroyed in some cosmic event. Humans cannot survive extremely long space journeys, while a superintelligence could. Only a superintelligence capable of making critical decisions on its own, and inventive enough to solve problems we cannot, could give the human race a second chance at survival :)

Indeed, it sounds like science fiction. But everything you can imagine is possible or will become feasible.

The Real Intelligence: Modelling the Human Brain

To create AGI beyond automation and content generation, such as in LLMs, we need to stop “reinventing the wheel.” It would also be helpful to look more closely into understanding the human brain to comprehend how intelligence and creativity develop. Only when we truly understand what constitutes the real human intelligence and how it functions in nature can we start modelling and improving it in AI.

What do you think? Would it be brilliant?

Conclusion: Comparing the Three Types of AI

Narrow AI is a domain-specific system in wide use today; General AI is a hypothetical human-level system that can reason across domains; and Superintelligence is a hypothetical system that exceeds human intelligence in every domain. Understanding these three types of AI and their capabilities is crucial for navigating the future of technology. While Narrow AI is already prevalent, the development of General AI and Superintelligence remains uncertain, but their potential impact is undeniable. As we continue to explore the possibilities of AI, it is essential to consider both the benefits and the risks associated with this transformative technology.

Thanks for reading my blog and subscribe (when you have not subscribed yet) to get new ideas and learn with me.

AI apps for Text

Try the following fantastic AI-powered applications.

I am affiliated with some of them (to support my blogging at no cost to you). I have also tried these apps myself, and I liked them.

Chatbase provides AI chatbots integration into websites.

CustomGPT.AI is a very accurate Retrieval-Augmented Generation tool that provides accurate answers using the latest ChatGPT to tackle the AI hallucination problem.

Flot.AI assists in writing, improving, paraphrasing, summarizing, explaining, and translating your text.

MindStudio.AI builds custom AI applications and automations without coding. Use the latest models from OpenAI, Anthropic, Google, Mistral, Meta, and more.

Originality.AI is very effecient plagiarism and AI content detection tool.