Definition
The Turing test is a procedure proposed by Alan Turing in his 1950 paper Computing Machinery and Intelligence. A human judge converses (via text) with two unseen participants — one human and one machine — and tries to determine which is which. If the machine can fool the judge consistently above chance, Turing argued, the machine should be considered intelligent. Turing called it the "imitation game."
Why it matters
How it works
A judge holds typed conversations with two unseen interlocutors — one human, one machine. The judge can ask anything. After a fixed time, the judge guesses which is which. Repeat with many judges and many machines. Aggregate scores measure how well the machine imitates. Variants include the "total Turing test" (with perception and action), restricted-domain tests, and reverse Turing tests (the machine judges). The test's exact rules and difficulty depend on the human level of the conversation, judge expertise, and time allowed.