HOW TO TELL YOUR FRIENDS FROM MACHINES

Abstract

WHEN a problem has been around as long as has the problem of minds and machines (or, for that matter, its twin the problem of' other minds') without either being very productive or being laid to rest, it is perhaps time to look beyond simple conceptual muddles and unclarities to deeper sources in philosophic practices and assumptions that enter undiscussed. To get the proper measure of the problem it may even be necessary to go beyond the stricter bounds of philosophy itself to look at the sources and role of certain philosophic views and problems in a wider social context, to provide what might be called a ' social pathology' of the problem. But though this may provide some understanding of the longevity of certain muddles, it raises, in its turn questions about the nature of philosophic problems that I do not see my way clear to tackling. I will confine myself to trying to bring out the basic muddle in the notion of 'machine intelligence', that is, the contradictions lying between the conditions of application of its two parts, and the slightly different muddle lying similarly at the centre of the now more popular notion of 'artificial intelligence'. I shall also make some suggestions about the social roots. Other extra-philosophic analyses have recently been offered. Geachl has turned to the notion of religious deviance in characterizing the belief in machine intelligence as 'idolatry'. It is not clear whether he intends this characterization as a contribution to our understanding of the attraction the notion seems to have for some people or whether it is meant simply as a piece of what might be called 'dissuasive description '-or, more baldly, ' name calling'. I do not share Geach's religious convictions, and perhaps because of that the characterization does not seem to me to be effective in either role. Without some pretty large assumptions about men's need to worship something-or-other, it hardly helps us to understand or explain the belief in intelligent machines, and as name-calling it would be least likely to sting just those people on whose heads it was being rained down. One justification for bringing in the medical term 'pathology' to describe the kind of account needed (though not the only justification) is apparent when one notices that a serious belief