The Turing Test of 1991

Abstract

Interactive computer programs and human participants competed in a Turing Test at the Boston Computer Museum last November in the first year of a competition to determine, ultimately, whether such programs can be indistinguishable from humans in dialogues. The test is named for the British mathematician and computer pioneer who proposed it in 1950. This paper describes the competition, its preparation, and problems which await resolution in future Turing Tests that may culminate in a $100,000 award. The 1991 test was “restricted” in its rules and procedures lest a full test disadvantage the computer programs too severely. The contest posed issues concerning dialogue domains, language processing, inclusion of cognitive tasks, and other features. Since a Turing Test could be interpreted as involving “thinking” and “intelligence” (though Turing had little use for such terms), future tests should intrigue human factors.