Back to Study Material Artificial Intelligence

Types of AI

AI systems are most commonly classified along two axes: capability and functionality. On the capability axis, researchers describe three levels. Narrow AI,...

Types of AI

Classifying by Capability

AI systems are most commonly classified along two axes: capability and functionality. On the capability axis, researchers describe three levels. Narrow AI, also called Artificial Narrow Intelligence or ANI, refers to systems designed and trained for one specific task. Every AI product available today, from virtual assistants to fraud detectors, falls into this category. These systems can outperform humans within their narrow domain but fail completely outside it.

Artificial General Intelligence, or AGI, refers to a hypothetical system that can perform any intellectual task a human can, switching between domains with the same flexibility a person does. No AGI system exists today, though it remains an active research goal. Artificial Superintelligence, or ASI, goes further, describing a system that surpasses human intelligence across every dimension. ASI remains firmly theoretical and is the subject of significant philosophical debate about safety and existential risk.

On the functionality axis, AI is classified as reactive machines, limited memory systems, theory of mind systems, and self-aware systems. Reactive machines, like IBM's Deep Blue chess engine, respond to inputs without any memory of past interactions. Limited memory systems, like self-driving cars, use recent observations to inform current decisions. Theory of mind and self-aware AI remain research concepts rather than deployed products.

A Concrete Example

A chess-playing program is a classic reactive machine. It evaluates the board state and selects the best move without remembering previous games or forming a model of its opponent's psychology. By contrast, a modern language model retains context within a conversation window, making it a limited memory system.

Exam Tip

Exam questions frequently ask students to classify a given system. Remember that every real AI system today is narrow AI. If a question describes a system that excels at one task but cannot generalise, that is ANI. The terms weak AI for narrow and strong AI for general intelligence are older synonyms you may also encounter in textbooks.