Types of Artificial Intelligence

Photo of author

By PeterLogan

Types of Artificial Intelligence

Artificial Intelligence’s main goal is to enable machines perform human-like functions. AI’s ability to reproduce human-like actions is the main way it is classified. Two types of AI can be classified, one based upon its ability to duplicate the human brain. Another type of classification is “Based On Functionality”. This classifies AI based upon their similarities to the human brain and their ability to think like humans. The second type of classification is much more popular in tech. It is “Based on Capabilities of AI vis-a – Human Intelligence”.

Different Types Of Artificial Intelligence

There are two types of AI that focus on Functionality & Capabilities.

Type 1 – Based on Functionality

Reactive Machine

These are the simplest form of Artificial Intelligence and also the oldest. They can mimic the human ability to react to different stimuli. This type of AI has no memory power, so they lack the capability to use previously gained information/experience to obtain better results. This type of AI is not capable of training themselves as well as the ones that we have today.

Exemple: Deep Blue (IBM’s chess-playing Supercomputer) is a perfect example of such machines. It is well-known for its defeat of Garry Kasparov, the international grandmaster in chess, in late 1990s. Deep Blue can recognize different pieces of the chessboard, and which moves they are. It can identify all possible legal moves for its opponent and itself. It selects the best option based on its options. These machines do not have memories so it can’t learn from previous moves.

Limited Theory

This type of AI, along with the ability of Reactive Machines, have memory capabilities so they can use past information/experience to make better future decisions. These are the most common applications that we see. These AI applications are easily trained with large quantities of training data that they keep in their memory.

Exemple: Limited Memory technology can be found in many self-driving vehicles. These data include GPS location, speed, obstructions size and nature, as well as other data that can be used to drive the car just like a person.

Theory of Mind

Theory of Mind is the next stage of AI. It has very little or no impact on our daily lives. These kinds of AI are typically in the “Work in Progress”, stage, and are often confined to research laboratories. These AIs will once they are developed will have an in-depth understanding of the human mind. It will include their needs, emotions, thought processes, and so on. Based on its understanding of Human minds, and their whims the AI can alter its response.

Example. Researcher Winston showed a prototype robot that can move down a narrow corridor with other robots from the opposite direction. The AI is able to foresee the movements of other robots and can make a turn right, left, or in any other direction so as not to collide with the incoming ones. Wilson says that the Robot decides how to move based on “common sense”.

Self-Aware AI

This is the final stage in AI. Its current existence can only be imagined and found in Science fiction movies. These AIs are capable of understanding and evoking human emotions, and can even create their own emotions. These types of AI may not be possible for many decades. These AI are the ones that AI skeptics, like Elon Musk, are cautious about. The reason is that once an AI is self-aware it can enter Self-Preservation mode. This means it may consider humans as a threat and might directly or indirectly seek to end humanity.

Type 2 – Based on Capabilities

Artificial Narrow Intelligence

This includes all AI applications we are familiar with. ANI refers to an AI system capable of performing narrowly defined tasks similar to humans. They are not capable of performing tasks it hasn’t been programmed to do, so they can’t perform an unusual task. According to the classification above, this system is a combination all-reactive and limited memory AI. These are the AI algorithms that we use to do complex Prediction Modelling in today’s world.

Artificial General Intelligence

AGI is able to learn, understand, perform and train functions much like a normal person. These systems will have multiple-functional capabilities that span across different domains. These systems will be more agile and can react and improvise as humans, even in the face of unprecedented circumstances. While there is no concrete example of AI of this type, there has been significant research.

Artificial Super Intelligence

Artificial Super Intelligence is the highest-level point in AI development. ASI will become the most powerful form ever of intelligence on Earth. Due to its extraordinary superior data processing, decision-making and memory abilities, ASI will be capable of performing all tasks more efficiently than humans. Some researchers fear that ASI could lead to the “Technological Singularity”. It’s a scenario where technology growth will be uncontrollable and result in a radical change in Human Civilization.

It’s difficult to see what the future holds for AI that is more sophisticated than we currently have. It is still far away from that stage. We are in the very beginning stages of advanced AI development. AI proponents can state that we are just scratching to uncover the true potential of AI. AI skeptics can only say that it is too early to start worrying about Technological Singularity.

Additional Resource: