What Are the 4 Types of Artificial Intelligence?
There are four types of artificial intelligence: reactive AI, limited memory AI, theory of mind, and self-aware AI. Only the first two actually exist in business today. The other two remain in the realm of research or speculation. Here is what each type means concretely for an executive.
Type 1: Reactive AI
This is the simplest form. It receives an input, produces an output. No memory. No learning. No context.
A concrete example: Netflix’s recommendation engine suggesting a film based on what you are watching right now. Or IBM’s Deep Blue chess system, which beat Kasparov in 1997 without ever remembering a previous game.
For a business, this type of AI is useful for repetitive, well-defined tasks: data filtering, anomaly detection in a supply chain, automatic document sorting.
Its limit is clear: it does not adapt. Change the context, it fails.
Type 2: Limited Memory AI
This is the dominant type today. It learns from historical data to improve future decisions.
This principle underpins the large language models currently being deployed in enterprise settings. The AI observes millions of examples, adjusts its parameters, and gradually becomes more accurate.
Maroc Cloud recently launched Gemini Enterprise in Morocco, with the stated objective of channelling the rise of AI in business and governing its use in a professional environment. This is a direct illustration of type 2 AI deployed at organisational scale. Morocco also ranks 66th globally among users of Claude, confirming that type 2 adoption is already a reality in the region.
For a CHRO or CEO, this is the type of AI that is changing processes today: CV analysis, staff turnover forecasting, report generation, decision support.
If you are wondering how to integrate this type of AI into your HR functions, my practical guide on AI in human resources details the operational use cases.
I have built a diagnostic framework to assess an organisation’s AI maturity across six dimensions. Download the Board Pack AI 2026 to structure your approach.
Type 3: Theory of Mind
This type does not yet exist outside research laboratories.
The concept: an AI capable of understanding the emotions, intentions, and beliefs of a human interlocutor. Not just responding to what is said, but understanding why it is said.
Prototypes exist in very limited contexts. Some social robots like Hanson Robotics’ Sophia simulate emotional reactions, but this is simulation, not genuine understanding.
For an executive, this type of AI is not a decision topic today. It is a topic for strategic monitoring.
Type 4: Self-Aware AI
This is the ultimate stage, often called artificial general intelligence or superintelligence. An AI that would have awareness of its own existence, its limits, its objectives.
It does not exist.
This concept attracts intense media coverage. These debates are legitimate within research circles, but they should not divert an executive from strategic prioritisation: types 1 and 2 are already in your systems or those of your competitors, and that is where concrete decisions are made.
What This Means for Your Organisation
Confusion between these four types is costly. Executives invest in tools believing they are buying an AI that “understands” their business, when they are actually buying a limited memory system trained on generic data.
The question is not “which AI is best?” in the abstract. It is: which type of AI matches which problem in my organisation?
For SMEs just getting started, this guide on choosing an AI solution asks the right questions before investing.
What I observe with my clients: most projects that fail do not fail because of the technology. They fail because nobody defined which type of AI was needed, for which process, with which data.
If you want to structure your approach before making an investment decision, request a diagnostic. We look together at where you stand and what makes sense for your context.
FAQ
What is the difference between weak AI and strong AI?
Weak AI is what exists today: systems designed for a specific task, however capable they may be. Strong AI, sometimes called artificial general intelligence, refers to an AI capable of reasoning on any subject like a human. It does not yet exist.
Is generative AI a separate type of AI?
No. Generative AI, such as GPT or Gemini, is an application of type 2: limited memory AI. It generates text, images, or code by drawing on patterns learned from massive datasets. It is not a fundamental type; it is a category of use.
What type of AI do large technology companies use?
Primarily type 2. Recommendation systems, fraud detection, search assistance: all of these rely on models trained on historical data. Some very simple systems still fall under type 1 for basic filtering tasks.
When can we expect type 3 or type 4 AI?
Nobody knows with certainty. The most serious researchers remain cautious and do not agree on a timeline. What is certain: it is not the subject of your next board meeting.