The Loebner Prize was a contest to implement the Turing test as a real, repeatable competition. Underwritten by Hugh Loebner, a New York manufacturer, and organized with the Cambridge Center for Behavioral Studies under Dr. Robert Epstein, it staged its first event on November 8, 1991, at the Computer Museum in Boston. Judges held typed conversations with a mix of hidden human “confederates” and computer programs, then ranked how human each one seemed. This account comes from Stuart Shieber, a Harvard computer scientist who took part in and analyzed the first competition.
The 1991 format was restricted: each program could pick a narrow conversation topic, which made the test much easier than Turing’s open-ended original. Of ten planned agents, six were programs; equipment failures cut the field to eight on the day. The winner was a chatbot by Joseph Weintraub that used crude, ELIZA-style tricks yet earned the highest average ranking - a result Shieber used to argue that the restricted contest tested showmanship more than intelligence. Loebner had earmarked a $100,000 grand prize and gold medal for the first program to pass a full, unrestricted Turing test; that grand prize was never awarded.
The Loebner Prize is historically important as the first sustained attempt to turn Turing’s thought experiment into an actual annual competition, and the debate it provoked - about whether fooling judges in a chat measures anything real - prefigured today’s arguments over how to evaluate large language models. For a general reader, it is a cautionary tale: a benchmark is only as good as its rules, and a poorly designed test rewards gaming rather than genuine ability.