Turing's imitation game by 2000 (1950 forecast)
Turing's closing prediction from *Computing Machinery and Intelligence*, storage scale and deception rate in the imitation game, evaluated against the year 2000.
Claim
By the year 2000, computers with about 10⁹ bits of storage will play the imitation game well enough that an average interrogator will not have more than 70 per cent chance of making the right identification (human vs. machine) after five minutes of questioning.
Follow-up article: turing-imitation-game-1950
Evaluation (as of 2000)
Storage: Roughly 10⁹ bits (about 125 MB) was ordinary for personal computers by 2000, this part of the forecast was conservative, not ambitious.
Imitation-game performance: No system in 2000 consistently met Turing’s behavioural bar in a fair, blind text setup: average interrogators could still often tell machine from human well above the 30% error rate implied by a ≤70% success rate. Early chatbots (e.g. ELIZA-style programs) fooled some users in narrow demos but did not sustain the effect under scrutiny.
Verdict: Partially true, hardware scale yes; the deception criterion, no (though later conversational systems revisit the same standard).