Artificial intelligence is used frequently to conjure up image of either cutting-edge advances to computer technology or a Terminator-style apocalypse, but what does it actually mean?
Artificial intelligence, or AI, is used to denote when computers or other man-made machines display intelligent behaviour, in contrast to simply responding to instructions set by a human. One of the key tenets of artificial intelligence is being able to learn from previous situations and being able to react appropriately based on that prior knowledge. A high-profile example of a machine displaying AI is IBM’s Deep Blue supercomputer, which was able to defeat World Champion Gary Kasparov at a game of chess in 1997. During all the contests between the two players, Deep Blue received no human input.
A more everyday example of AI comes in the form of Apple’s personal assistant Siri, which evaluates previous queries and requests to return results that are personalised to the user.
One of the leading tests to determine whether a machine is exhibiting “intelligence,” is the Turing Test, devised by British mathematician Alan Turing. The Turing Test involves a human judge alongside a human and computer participant. Upon asking the participants a series of questions, if the judge is unable to distinguish between the human and computer, the computer is deemed to be “thinking.”
The Turing Test does have its critics, but despite being devised more than 60 years ago it is still used as a benchmark for AI testing.
Of course, before a computer is able to take the Turing Test or any other form of AI assessment, humans face the challenge of devising an intelligent machine in the first place. Reason, problem solving, knowledge and learning are all difficult functions to replicate despite all being, for the most part, logic-based. When scientists then try to imbue machines with emotional intelligence or other human traits like impulse and creativity, the problems become even greater.
AI research is a varied discipline, but successes have been made at replicating logical intelligence. Primarily this involves gathering information through human input or sensors, which the computer can then analyse, before choosing an outcome. The previously mentioned Deep Blue is an example of this form of artificial intelligence.
In fact, chess computers highlight that AI has a long way to go before it can fully replicate the thought process of a human being. Instead researchers often have better results focusing on a specific area of intelligence or skill, such as playing chess. IBM’s supercomputer Watson used its four terabytes of disk storage and 16 terabytes of RAM to defeat two former contestants on US quiz show Jeopardy, but still struggled with shorter clues and was unable to respond to the other contestants’ answers. Currently, even the most advanced artificial intelligence machines are plagued by limitations.
For now, artificial intelligence of the kind that has inspired literature and movies remains firmly rooted in science fiction, but that is not to discredit the research being carried out. Optimistic outlooks suggest that the development of AI could bring tremendous advances to mankind and ultimately it may one day be able to replicate human intelligence, or perhaps even surpass it.