What types of A.I. exist and how do we classify them?
There are three main types of artificial intelligence that we encounter in A.I. literature: narrow A.I., general A.I., and super A.I. These three types of A.I. differ mainly in terms of their capabilities and flexibility. But what do these terms mean and what can each of them do?
Artificial Narrow Intelligence is a type of A.I. designed to solve a very narrow set of specific problems. For example, a Roomba can vacuum your floors, Amazon Alexa can respond to basic voice commands, and Google’s Waymo is a self-driving car. In each case, the agent is performing a small set of tasks in a narrow domain of knowledge.
Artificial Narrow Intelligence is more commonly known as “Narrow A.I.” or often abbreviated as “A.N.I.”. In addition, it may also be referred to as “Applied A.I.” or “Weak A.I.”.
This is a type of A.I. that exists today, is growing quickly, and is rapidly changing our world.
Artificial General Intelligence is a futuristic type of general-purpose intelligence that can solve a wide variety of problems. This is commonly associated with sentient robots in books, TV, and film, like: R2-D2 from StarWars, Data from Star Trek, or HAL 9000. In each case, the agent is able to flexibly perform a much wider set of tasks across various domains of knowledge.
Artificial General Intelligence is more commonly known as “General A.I.” or often abbreviated as “A.G.I.”. In addition, it is sometimes referred to as “Strong A.I.” or “Full A.I.”.
This type of AI does not yet exist and is still likely many years (if not decades) from becoming a reality. However, this is the general direction of A.I. and the goal of many A.I. research projects.
Artificial Super Intelligence is a type of futuristic A.I. that is smarter than all human (or natural) intelligence. These are typically the supervillains in Sci-Fi movies. For example, Skynet from The Terminator, Ultron from the Avengers, and The Architect from The Matrix. In each case, the agent is able to perform super-human tasks that exceed the abilities of any single human being.
Once we achieve artificial general intelligence, it is believed that artificial super intelligence will emerge soon after. This is because general A.I. will be able to recursively self-improve and it is not subject to the biological limitations of organic brains. If it ever comes into existence, Super A.I. would likely vastly exceed all forms of human intelligence on Earth (combined).
All three of these types of A.I. are interesting topics to discuss for various reasons. However, unless your planning to start a career in advanced A.I. research, for most of you, narrow A.I. is the type of A.I. you should be investing your time in today.
It’s A.I. that is designed to solve a very narrow set of specific problems but will likely revolutionize our world in more ways than we can imagine over the next few decades.
To learn more, please check out my latest online course: Artificial Intelligence: The Big Picture.