Your Friends and Your Machines

Abstract

Mr. Robinson's article 'How to Tell Your Friends from Machines' is an interesting exercise in standing John Wisdom's techniques on their heads. Wisdom used to argue that certain philosophical beliefs were defended in ways similar to neurotic beliefs in clinical experience. Mr. Robinson seems to be arguing that belief in the existence, or possibility, of intelligent machines is neurotic or even psychotic, therefore it is ipsofacto not a philosophical belief. As a general form of argument this will not, I fear, serve Mr. Robinson's overall purpose, for Wisdom's more strongly entrenched arguments suggest that, if an associated neurosis disqualifies a belief as philosophical, then a great number of other apparently philosophical beliefs, such as the nonexistence of material substance, will have to go as well, and the philosophical cupboard may end up looking rather barer than Mr. Robinson intended. Mr. Robinson's case seems to me to contain two specific defects: one of fact and one of principle. He says that '. . . our society's reaction to behaviour that manifests a genuine belief in the intelligence of machines is quick and extreme' (p. 505). It is perhaps unfair to make an ad hominem reply that involves professors, by pointing out that many such believers are, on the contrary, elevated to chairs of psychology, computer science and machine intelligence, so let us stick to the man in the street. Mr. Robinson is simply wrong, for quite ordinary people already discuss, with all appearance of serious belief, whether or not a given trick will outwit a credit card company's computer. Mr. Robinson is too late, for that is already an established and well understood way of speaking. The error of principle is that he seeks to single out a philosophical problem, that of minds and machines, and sometimes that of 'other minds' as well, on grounds which do not, in fact, distinguish it from other familiar philosophical problems. 'The problem of machine intelligence', he writes, 'is not a real problem for us in the sense that nothing hangs on its solution' (p. 506). But in the terms of his argument, where what can hang are 'legal, moral and social or diagnostic decisions', what hangs on the solution of any philosophical problem? That line of attack can only be on philosophy as a whole, not on a sub-problem within it. In spite of these two defects, Mr. Robinson's paper is worth discussing, and on the ground he himself chooses, that of 'social pathology of beliefs about machines', rather than on the well-trodden ones he avoids so carefully, where the question 'Are there intelligent machines?' is tackled head on. The social point on which I think Mr. Robinson is mistaken, and importantly so, is that not only is there a widespread attribution of intelli