thrift@relief.UUCP (10/20/87)
In Turing's original paper on whether machines can think, he considers opinions contrary to his own. These include what he calls the `Theological Objection' (machines don't have souls and therfore can't think), the `Mathematical Objection' (Godel's Theorem, etc.), and others. He provides arguments countering all these objections, but I was surprised when I first read this paper many years ago to see his statement and argument against the last objection: `The Argument from Extra-Sensory Perception'. He states that "the statistical evidence, at least for telepathy, is overwhelming". Even more surprising than his belief (at the time) in the validity of the evidence was his counterargument. The objection raised was that a difference could be detected between a machine `guessing' and a person (with ESP) `guessing' the suit of a card in the interrogator's hand (the machine would be correct on average 25% of the time, the person statistically better). But Turing sees an opening. The machine will probably have a random number generator (to make the `guesses'), but a random number generator "subject to the psycho-kinetic powers of the interrogator" and hence "might cause the machine" to match the telepathic powers of the person. He suggests the test should be tightened by putting the competitors in a "telepathy-proof" room. Philip Thrift