Artificial intelligence includes many different technologies and systems. Some AI tools are designed for a specific task, while others can perform a wider range of activities.
Because AI is such a broad field, it is often classified in different ways.
Two common approaches are to classify AI by its capabilities and by its functionality.
This beginner’s guide explains these main types and how they relate to the AI systems used today.
AI Types Based on Capabilities
A common way to classify artificial intelligence is by how broadly a system can perform intellectual tasks.
The three main categories are:
- Narrow AI
- Artificial General Intelligence
- Artificial Superintelligence
Narrow AI
Narrow AI, sometimes called weak AI, is designed or trained to perform specific types of tasks.
Most AI systems used today belong to this category.
Examples include:
- Recommendation systems
- Image recognition
- Spam filters
- Translation tools
- Navigation applications
- Fraud detection
- AI chatbots and assistants
A narrow AI system can be very capable within its intended area without possessing general human intelligence.
Modern large language models can perform many language-related tasks, but this does not automatically mean they have human-level general intelligence.
Artificial General Intelligence
Artificial General Intelligence (AGI) refers to the idea of an AI system that could perform intellectual tasks across many different areas with broad and flexible abilities comparable to human general intelligence.
Such a system would theoretically be able to learn new tasks, apply knowledge between different domains, and adapt to unfamiliar situations.
AGI remains a research goal and should not be assumed to describe ordinary AI systems available today.
Artificial Superintelligence
Artificial Superintelligence (ASI) is a hypothetical form of AI that would exceed human intellectual capabilities across a broad range of areas.
This could include reasoning, planning, scientific discovery, creativity, and problem-solving.
Artificial superintelligence does not currently exist and is mainly discussed in theoretical research and debates about the future of AI.
AI Types Based on Functionality
AI can also be classified according to how a system processes information and interacts with its environment.
A commonly discussed classification includes:
- Reactive Machines
- Limited-Memory AI
- Theory-of-Mind AI
- Self-Aware AI
Reactive Machines
Reactive Machines respond to current inputs without relying on a stored history of previous experiences.
A simple example is an AI system designed to play a board game by analysing the current situation and selecting its next move.
Reactive systems can perform particular tasks effectively, but they do not build a general memory of previous experiences.
Limited-Memory AI
Limited-Memory AI uses past or recent data to help produce predictions or decisions.
Many practical AI systems can be understood within this broad category.
Examples include:
- Autonomous driving systems
- Fraud detection
- Recommendation systems
- Predictive maintenance
- Machine-learning predictions
Modern machine learning and deep learning rely heavily on learning patterns from previous data.
Theory-of-Mind AI
Theory-of-Mind AI refers to the idea of systems that could understand human beliefs, emotions, intentions, and social behaviour in a deeper way.
Current AI can recognise some conversational or emotional patterns, but this should not automatically be interpreted as genuine human-like social understanding.
Theory-of-mind AI therefore remains largely a research concept.
Self-Aware AI
Self-Aware AI is the theoretical idea of artificial intelligence possessing awareness of itself and its own internal state.
There is currently no established self-aware AI system.
A system that produces human-like conversation does not necessarily possess consciousness or self-awareness.
Where Does Generative AI Fit?
Generative AI is not a separate intelligence level such as narrow AI or AGI.
Instead, it describes AI systems designed to create new content, including:
- Text
- Images
- Audio
- Video
- Computer code
Many current generative AI systems are still best understood as forms of narrow AI, even when they can perform many related tasks.
Large language models, for example, can answer questions, summarise information, assist with writing, and generate code without necessarily possessing general human intelligence.
Which Type of AI Do We Use Today?
Most AI tools people use today are forms of narrow AI designed or trained for particular tasks.
This includes recommendation systems, translation tools, image recognition, fraud detection, chatbots, and many Generative AI applications.
Even when a modern AI system can perform several different tasks, this does not automatically make it AGI.
Why These Classifications Matter
Understanding the types of artificial intelligence helps explain what current AI systems can and cannot do.
It also helps distinguish between terms such as AI, machine learning, generative AI, AGI, and other advanced concepts.
These classifications are useful when discussing:
- AI capabilities
- Technical limitations
- Potential risks
- Human oversight
- Ethics and responsibility
- Future development
It is especially important to separate AI technologies that exist today from theoretical systems that may or may not be developed in the future.
Final Thoughts
Artificial intelligence can be classified in different ways depending on the purpose of the classification.
By capability, AI is commonly discussed as narrow AI, artificial general intelligence, and artificial superintelligence.
By functionality, common conceptual categories include reactive machines, limited-memory AI, theory-of-mind AI, and self-aware AI.
Most AI systems currently used in everyday life are forms of narrow AI designed or trained for particular tasks.
Understanding these distinctions makes it easier to evaluate new AI technologies realistically and understand where current systems fit within the wider field of artificial intelligence.
At AIWiseUp, we’ll continue exploring artificial intelligence in practical and understandable terms.
Wise up to AI. Learn it. Use it. Grow with it.
References and Further Reading
Russell, S.J. and Norvig, P. (2021) Artificial Intelligence: A Modern Approach. 4th edn. Pearson.
Artificial Intelligence: A Modern Approach – Pearson
Goodfellow, I., Bengio, Y. and Courville, A. (2016) Deep Learning. MIT Press.
Deep Learning – Official Book Website
Kaplan, A. and Haenlein, M. (2019) ‘Siri, Siri, in my hand: Who’s the fairest in the land? On the interpretations, illustrations, and implications of artificial intelligence’, Business Horizons, 62(1), pp. 15–25.
Siri, Siri, in my hand – Business Horizons DOI
Legg, S. and Hutter, M. (2007) ‘Universal Intelligence: A Definition of Machine Intelligence’, Minds and Machines, 17, pp. 391–444.