Britannica Sources used in this answer

How does the Turing test try to evaluate whether a computer can think?

Created from Britannica articles using AI. AI can make mistakes, so verify using Britannica articles.

Artificial intelligence — artificial intelligence Image generated by the Stable Diffusion model from the prompt “the ability of a digital computer or computer-controlled robot to perform tasks commonly associated with intelligent beings,” which is the definition of artificial intelligence in the *Encyclopædia Britannica* article on the subject. Stable Diffusion is trained on a large set of images paired with textual descriptions and uses natural language processing to generate an image.

The Turing test is a practical, behavioral approach proposed by Alan Turing in 1950 to sidestep the philosophical difficulty of defining "thinking" in machines. Instead of seeking an objective definition of intelligence, the test evaluates whether a computer can act, react, and interact in a way that is indistinguishable from a human.12

How the Test Works

The test involves three participants communicating via keyboard and display screen:2
* The Interrogator: A human who asks wide-ranging questions to determine which participant is the computer and which is the human "foil".2
* The Computer: Attempts to convince the interrogator that it is the human, often by mimicking human-like errors or delays.2
* The Human Foil: Assists the interrogator by providing natural, human responses.2

If the interrogator cannot reliably distinguish the computer from the human after a series of tests, the computer is considered to have demonstrated thinking ability.13

Criticisms

Critics argue that the test is an insufficient measure of intelligence. The Chinese room argument suggests a machine could pass by merely following instructions without understanding, while others note that a machine could theoretically pass by using a pre-programmed table of responses.41