Turing test
Test for machine intelligence. Alan Turing, 1950 paper "Computing Machinery and Intelligence". Original name: "the imitation game". A human judge chats with a human + machine → if they can't tell apart, the machine has passed.
Philosophically controversial — Searle's Chinese Room (1980) argues passing ≠ understanding. The Loebner Prize (1990–2019) was an annual contest. Chatbots like ELIZA (1966) fooled people early without "understanding". Modern LLMs (GPT-4+, Claude) routinely pass simple Turing tests, but no serious researcher claims they "think". 2024–25 focus shifted to AGI benchmarks (ARC-AGI) measuring generalization ability instead.