On the Validity of the Turing Test