Artificial Intelligence is divided into two different stages based on capabilities and functionalities of the machines are as follows:
Type-1 (Based on capabilities)
- Artificial Narrow Intelligence (ANI)
- Artificial General Intelligence
- Artificial Super Intelligence
Type-2 (Based on Functionalities)
- Reactive Machines
- Limited Memory
- Theory of mind
- 1 Artificial Intelligence Type-1: (Based on Capabilities of machines)
- 2 Artificial Intelligence Type-2: (Based on Functionalities of machines)
Artificial Intelligence Type-1: (Based on Capabilities of machines)
Artificial Narrow Intelligence
Artificial Narrow Intelligence is also known as weak AI. It is applied only for some specific tasks. Most of the currently existing systems that claim to use AI are actually operating as weak AI focused on a narrowly defined specific problem.
Example– Alexa is the best example of Artificial Narrow Intelligence or weak AI because it operates within an unlimited pre-defined range of functions. It does not have any genuine intelligence or self-awareness.
Some other examples of weak AI are the Google search engine, Sophia the humanoid, self-driving cars, and famous AlphaGo.
Artificial General Intelligence (AGI)
Artificial General Intelligence (AGI) is also known as strong AI. It involves a machine that passes the ability to perform any intellectual task that a human being can perform.
Machines don’t possess human-like abilities. However they have a very strong processing unit that can perform very high-level computations, but they are not yet capable of doing simple and most reasonable things that a human being can perform. A machine can review millions of documents in a few seconds or minutes only but if you ask a machine to walk up to your living room and switch on the fan, a machine will take forever to learn that because machines don’t have a reasonable way of thinking.
So we can conclude, machines have a strong processing unit but they can’t think like a human being. There is no such machine developed yet that is fully strong AI.
Example– AlphaGo Zero has defeated AlphaGo in the game of Go.
Artificial Super Intelligence (ASI)
Artificial Super Intelligence (ASI) is a term referring to the time when the capabilities of computers will surpass humans. However, it is a hypothetical concept as depicted in movies or science fiction books where machines have taken over the world. Movies like Terminator depict Artificial Super Intelligence.
These do not exist yet in the real world but there are lots of people who speculate that ASI will take over the world by the year 2040.
Note- Narrow intelligence is the only thing that is existing till now which means we have only weak AI or weak Artificial Intelligence in the real world. The main difference among above types is Narrow AI is dedicated to one task, General AI can perform like a human being and Super AI can think intelligently than a human being.
Artificial Intelligence Type-2: (Based on Functionalities of machines)
Purely reactive machines are the most common types of AI that don’t have the ability to form memories and use pass experiences for current decisions. IBM chess-playing supercomputer and Deep blue are the perfect examples of Reactive machines, which defeated the chess grandmaster Garry Kasparov in early 1990.
Deep blue is a technique that is used to identify the pieces on the chessboard and each new move. It can predict for the next movement of the opponent and the most optimal movement from all available possible movements. Deep learning can’t able to memorize the past moves or does not have any past memory. It ignores the rarely used chess-specific rules against repeating the same technique three times.
Similarly, Google’s AplhaGo has beaten top human experts in Go. It also can’t evaluate the potential future moves. But it can analyze better than Deep blue by using Neural Network to evaluate the game development. These machines can function only for the assigned tasks and can be easily fooled.
Apart from Reactive machines, these machines have the ability to use past experience for current decisions. Self-driving cars have proven it already because it is not easy to predict the speed and direction of other cars and object randomly. These predictions through self-driving cars added as a programmed representation including traffic signaling, lane marking, the curve in the road, and other traffic elements. These machines help to decide when to change the lane and overtake to the other cars without any accident.
These pieces of information are instantaneous about the past as this information never saved in the car’s database of experience from where it can learn in the future. So a system that can use past experience and also can handle new situations lies in this category.
Theory of Mind
These types of AI machines are performing as a bridge among Reactive machines, Limited memory, and the machines that will build in the future.