The Mismeasure of Machines

摘要

We reply to Hector Levesque's critique of the Turing Test and his proposal for "a new type of Turing Test", a Winograd Schema Test. We question whether the role of deception in the Turing Test is, as Levesque asserts, "a serious problem". We argue that the Levesque Test specifies the nature of intelligence in a way that Turing wished to avoid. We conclude that the Turing Test appeals to the collective judgment of humankind, a judgment that, in the case of AI, has yet to be rendered.