Alan Turing – “Computing Machinery and Intellegence”
In this article, Alan Turing discusses whether machines can think. He makes thinking equivalent with the machine being able to simulate a human in an experiment where an interrogator has to find out which one of the respondents is human or machine by asking them questions. If the machine is able to pass this test, it is said to show intelligence. As the text was published in 1950, Turing has to do some explanation of what he means by machines (digital computers), their functionality (storage, executive unit/ programming, and control) and abilities. An important quality of the digital computers is that they can be programmed to have a random functionality and that they can imitate state machines (which have a finite number of possibilities or states as in and output), giving them a universal character. He predicts the future development of their speed and storage capacity as such that by 2000, it should have been possible to make a digital computer that can pass the test, but actually the first one that is said to have passed the Turing test dates back to 1966: the chatbot ELIZA.
In the second part of the article, Turing goes on to dismiss counterarguments to his statement that machines can think (under his conditions of what thinking means). He dismisses the theological and intellectual objection that thinking is only reserved for the powerful human and also the consciousness argument that computers cannot feel is said to be solipsist and from a human point of view. The mathematical and various disabilities objection focus on things that the computer cannot do, such as giving answers to critical questions and answers based on induction or rules of behaviour, can be dismissed because this does not necessarily give away the computer in the test (as it can give a prediction of the answer or one based on randomness) as also humans make mistakes.
Turing ends his article by focussing on the learning abilities of computers. They can program themselves by trial and failure. Turing thinks that it might be easier to aim at creating a “child machine” than an “adult machine” because the first will like a child have all the equipment to learn and can evolve by itself to become more capable and have more data or rules stored. Moving from his first statement that in fifty years computer will be able to pass his test, he predicts that “machines will eventually compete with men in all purely intellectual fields.” Whether in abstract activities as chess or in understanding and speaking a language.
Questions: What do you think of Turing’s comparisons between machines and humans? He writes that it is unfair to expect of one of them to be like the other in all aspects (“We do not wish to penalise the machine for its inability to shine in beauty competitions, nor to penalise a man for losing in a race against an aeroplane.”), but what do you think of the choices he makes of qualitites in which computers should be able to compete with humans?