In 1950 Alan Turing sidestepped the unanswerable "can machines think?" with a practical game: if a human judge, chatting blindly via text, cannot reliably tell a machine from a person, the machine passes. Behavior, not metaphysics, becomes the test.
What it does and doesn't settle
- It reframes thinking as something you demonstrate, dodging the hard problem entirely.
- Critics say passing only proves good imitation — the heart of The Chinese Room objection. Fluent output may be simulation, not understanding.
Modern LLMs routinely fool people in short exchanges. Did they pass the test — or did the test never measure what we hoped it did?
The Turing Test's real legacy may be showing how slippery "thinking" is to define once a machine starts doing it.
Related: The Chinese Room · The Hard Problem of Consciousness · Mary's Room and Philosophical Zombies