An interrogator chats with two hidden players, one human, one machine, and decides which is which.
From the chapter: Chapter 1: What Is AI?
Glossary: turing machine
Transcript
The year is 1950. Alan Turing replaces the question "can machines think" with a game.
Three players. An interrogator on one side of a wall. On the other side, a human and a machine. The interrogator can only see typed messages.
The interrogator types a question. Both the human and the machine answer. The interrogator reads both, weighs them, and types another.
The aim, says Turing, is for the interrogator to decide which respondent is the human. The aim of the machine is to be mistaken for the human.
If, after enough conversation, the interrogator does no better than chance, the machine has passed the test.
Notice what the test sidesteps. It does not ask whether the machine has consciousness. It does not ask whether it understands. It asks only whether its behaviour, in language, is indistinguishable from a person's.
Modern critics say large language models pass this test routinely, and that the test was always too weak. Modern advocates say it is exactly as informative as Turing intended, no more and no less.
Either way, this is the moment AI becomes an empirical science.