In a study spanning 2025 and 2026, judges chatting through a text window identified an AI model as human 73 percent of the time, a higher rate than the real people in the same experiment achieved. This episode asks how that happened by going back to 1950, when Alan Turing, working at the University of Manchester, published Computing Machinery and Intelligence. He thought the question of whether machines can think was too vague to answer, so he replaced it with a parlor game.

In the original imitation game, a man tries to convince a judge by typewritten notes that he is a woman. Turing swapped the man for a computer. The episode connects that premise to Turing’s own life as a gay man forced to pass in mid-century Britain, then follows the test through ELIZA, PARRY, John Searle’s Chinese room, and the Loebner Prize, where programs won by faking typos. It ends with researchers who now see the test as a trap that rewards replacing people.

• Joseph Weizenbaum’s 1966 program ELIZA imitated a psychotherapist by turning keywords into questions, and he was disturbed by how readily people confided in it.
• Kenneth Colby’s 1972 program PARRY played a paranoid schizophrenic. Psychiatrists shown transcripts could tell it from real patients only about 52 percent of the time.
• Searle’s 1980 Chinese room argues that following rules to shuffle symbols is not understanding. The systems reply says the room as a whole does understand.
• Eugene Goostman, a 2001 chatbot posing as a 13-year-old boy from Odessa, Ukraine, fooled a third of the judges in a major competition.
• CAPTCHA stands for Completely Automated Public Turing Test to Tell Computers and Humans Apart: a reverse test in which the machine interrogates the person.