Watch and find out the history behind the burning question and development of The Turing Test!
Transcript:
In the world of artificial intelligence, there exists a burning question that everyone seeks an answer to — can machines think? Does it remind you of the Westworld TV show? Dolores, an artificial intelligence robot, who appears to be capable of developing consciousness, making her own decisions and deviating from the programmed script. Makes you think… how do we know whether machines can think?
Alan Turing, a British computer scientist and mathematician, who was well known for developing an algorithm that cracked German codes during the World War Two. He famously opened his 1950 paper with a question, “can machines think?” He developed a test, called The Turing Test or The Imitation Game, that tested a machine’s ability to exhibit intelligent behavior equivalent to or indistinguishable from a human. Intelligent behavior is defined as ability to demonstrate human intelligence in thoughts, words, or actions.
What does the game look like? There are three rooms, each connected via computer screen and keyboard. There sits a human in one room, a machine in the another, and a “judge” (a human) in the third room. The judge’s job is to ask questions to both participants and decide which of the two participants talking to him through the screen is the human. The human will attempt to help the judge in making the right decision by offering evidences. The computer’s job is to trick the judge and counteract the human’s claims in the hope the judge will erroneously identify the computer as the human. Turing proposed that if a judge is less than 50% accurate in choosing either human or computer, then the computer passed the test. It is thus defined as demonstrating human intelligence and therefore can “think.”
For forty years after Turing proposed his test in 1950, nobody has created a machine that is able to pass the test so the Loebner Prize was created in 1990. It is an annual competition where it will award small monetary amounts to computer programs considered by the judges to be the most human-like. There is $100,000 reward for the first program that judges cannot distinguish a computer from a real human in the Turing Test with textual, auditory, and visual components (with less than 50% accuracy) and if a program achieves it, the annual competition will end. So far, some people have created programs that passed the textual component; but, nobody has passed all three components.
The Turing Test and the Loebner Prize are highly controversial topics in the field of artificial intelligence. Some support the test as a measure of intelligence in machines. Some feel that the Turing Test is a poor test of intelligence; it encourages trickery not intelligent behavior and that many intelligent systems would fail this test.
Nevertheless, the question remains… can machines think? If we successfully created a Westworld, will Dorothy pass The Turing Test? Maybe we will know 100 years from now.
