Everyone's a critic: Memory models and uses for an artificial turing judge

Abstract

The Turing test was originally conceived by Alan Turing [20] to determine if a machine had achieved human-level intelligence. Although no longer taken as a comprehensive measure of human intelligence, passing the Turing test remains an interesting challenge as evidenced by the still unclaimed Loebner prize[7], a high profile prize for the first Alto pass a Turing style test. In this paper, we sketch the development of an artificial "Turing judge" capable of critically evaluating the likelihood that a stream of discourse was generated by a human or a computer. The knowledge our judge uses to make the assessment comes from a model of human lexical semantic memory known as latent semantic analysis[9]. We provide empirical evidence that our implemented judge is capable of distinguishing between human and computer generated language from the Loebner Turing test competition with a degree of success similar to human judges.