What does the "Turing test" measure?

The Turing test, proposed by Alan Turing in 1950, asks whether a machine can carry on a conversation convincing enough that a human judge cannot reliably tell whether they are talking to a person or a computer. Turing framed this as a practical stand-in for the harder philosophical question of whether machines can "think," sidestepping definitions of consciousness in favor of an observable behavior anyone could test.

It has nothing to do with calculation speed, which is simply a measure of raw processing performance and existed long before AI conversation was a goal. It also has no connection to battery life, which concerns power engineering rather than intelligence or conversation.

Since Turing's proposal, plenty of chatbots have briefly fooled some judges in short conversations, but many researchers argue this says more about how easily people can be misled in casual small talk than about genuine machine understanding.