Artificial intelligence is the ability of machines to manifest human abilities – such as reasoning or learning (you can read more about what artificial intelligence is here).
Artificial intelligence can be divided into different types. In this article, we outline two main types of AI categorization – based on the capabilities and functionality of artificial intelligence.
Classification of AI according to artificial intelligence capabilities
In this category, there are three types of artificial intelligence: narrow, general, and superintelligence.
So far, only narrow artificial intelligence has been created, but scientists are getting closer to developing general AI.
Artificial Narrow Intelligence (ANI) / Weak AI / Narrow AI
Artificial intelligence of this type is designed to perform single tasks – such as speech recognition, facial recognition, driving. It makes up most of the artificial intelligence around us. The machines appear intelligent, but work within a set of rules and limits. Such AI does not mimic human intelligence, but merely simulates human behavior.
Examples of the use of narrow AI:
- Chatbots;
- Voice assistants;
- Facial recognition systems;
- Self-driving cars.
Artificial General Intelligence (AGI) / Strong AI / Deep AI
These are machines that have intelligence equivalent to humans, are capable of learning, and can use their intelligence to solve any problem. Such artificial intelligence can think, understand and act in a way that is indistinguishable from humans.
Scientists have not yet created such an artificial intelligence because it would require the creation of a conscious machine with cognitive abilities.
Artificial Superintelligence (ASI)
Artificial superintelligence not only mimics or understands human intelligence and behavior, but is self-aware and surpasses the capabilities of human intelligence.
The concept of artificial superintelligence assumes that AI will evolve to not only understand human feelings, but also to evoke its own emotions, needs, beliefs, and desires.
ASI would have a larger memory and faster ability to process and analyze data.
Classification of AI according to functionality of artificial intelligence
There is another subdivision of artificial intelligence proposed by AI researcher Arend Hintze.
Reactive machines
These machines are created to solve a specific problem, have no memory, and therefore are unable to learn. The example is Deep Blue, the IBM chess program, which beat Garry Kasparov in the 1990s.
Machines with limited memory
These machines use information from databases, learn basic information from the environment. Some of the decision-making functions in self-driving cars are designed this way.
Machines with theory of mind
They are still under development – such AI is capable of understanding human emotions and making decisions based on that.
Machines with self-awareness
These machines are able to think, perceive and act like a human and has not yet been developed.
The idea behind artificial intelligence is that we can make computers that are capable of learning for themselves.
This is how it works: take an example such as learning how to play chess or driving a car, both tasks involve many rules and procedures which must be followed in order for success; however, there’s no reason why we couldn’t create an algorithm that could do this in its own way – without needing human assistance at all and even do better than humans. Thus was born the idea of superintelligence.
As we have seen in recent years, AI is going to be the future of our world and maybe even in a form that surpasses us in every way. With this in mind, it has become more important than ever before for governments to prepare themselves and their citizens for the challenges that lie ahead.
Autor: Agata Konieczna
Credits:
Robot icons created by Freepik – Flaticon
Capability icons created by monkik – Flaticon
Configuration icons created by Freepik – Flaticon
