The Turing Test
The Turing Test, in artificial intelligence, is a test to determine whether a machine can ‘think’, or behave, like a human.
It was named after its founder, Alan Turing, in 1950, in his paper titled ‘Computer Machinery and Intelligence’ at the University of Manchester.
The way it works is simple — it has a similar concept to the Imitation Game.
The Imitation Game
The original Imitation Game involves three participants — one male, one female, and one (male or female) judge — placed in different rooms. The only way they can communicate with each other is via screens and keyboards.
The male tries to persuade the judge that he is the female. The judge’s job is to distinguish who is male and who is female.
The Turing Test
In the Turing Test, the three participants are: one machine, one human, and one human judge. The judge’s job is to distinguish between the machine and the human in five minutes, by asking both of them a series of questions.
If the judge is unable to accurately distinguish human and machine based on their responses, the machine passes the Turing Test and is said to have ‘artificial intelligence’.
This simple test did its job well back in the day. It provided something — a measurement that allowed people to start discussions on the philosophy of artificial intelligence.
Without getting too philosophical, Turing’s 1950 paper replaced the popular question ‘Can machines think?’ with his version: ‘Can machines do what humans do?’
He believed ‘Can machines think?’ was a meaningless question to ask. His alternative question distinguishes between the physical and intellectual parts of the human.
But many philosophical considerations and loopholes soon surfaced. Thus, it is not a one-size-replaces-all test. Since then, there are many variations as well as alternative tests, which will be explored in future posts.
Programs like ELIZA might pass the Turing Test simply by manipulating the question, without actually comprehending or contextualising the question. The illusion of consciousness is enough to fool the judge into thinking it is human.
Moreover, the assumption behind his argument is that if machines are able to imitate human behaviour, then it means they are intelligent.
The original question, “Can machines think?” I believe to be too meaningless to deserve discussion. Nevertheless I believe that at the end of the century the use of words and general educated opinion will have altered so much that one will be able to speak of machines thinking without expecting to be contradicted.
However, who are we to say that human intelligence is superior to other forms of intelligence? How do we define ‘thinking’ — is machine thinking supposed to be equivalent to human thinking?
At the end of the day, there are many researchers who believe that passing the Turing Test should not be the end-goal. Is it really meaningful to measure machine intelligence based on human intelligence, and is it really necessary to have a machine imitate a human in order to achieve basic goals?
Nevertheless, Turing’s paper was one of the first to touch deeply on machine thinking and he has greatly influenced the field of artificial intelligence research.
Thanks for reading. Have a great day!