So Turing suggested replacing the question with the imitation game, which fixes certain variables in a rules-based scenario that is easily implementable and controllable. Suppose that A and B are a human being and a computer, but you do not know which is which. You can ask A and B any question simultaneously, but they are in another place and you can only interact with them by e-mail (or, in Turing’s day, by teleprinter). If after a reasonable amount of time you cannot tell which is the human and which the computer, then the computer has passed the test — that is, the computer is at least as good as the human in providing answers to the questions you asked. Turing’s test is based on a weaker version of Leibniz’s law of the identity of indiscernibles: if, everything else being equal, significant differences between A’s and B’s answers are indiscernible, then A and B are interchangeable. Given the same input of questions, the output of answers a human and a computer can generate are such that the differences between the two are insufficient for the purposes of unmistakable recognition.
Sunday, May 21, 2017
Leibniz’s law of the identity of indiscernibles and the Turing Test
Why Information Matters by Luciano Floridi.