Book review: Foundation of Artificial Intelligence by David Kirsh (ed.) (Cambridge, MA: MIT Press)
Ian E. Pratt-Hartmann · ACM SIGART Bulletin · 1993
Since the birth of AI in the 1950's, it has seemed to many in the field to be natural, even unavoidable, to characterize the states of intelligent systems in terms of what those states represent. Thus, programs are regularly said to know various facts or to have various goals. And researchers regularly attempt to write programs which can reason with that knowledge and those goals---programs, that is, which can perform certain sorts of inferences, devise plans in certain domains, generate hypotheses to explain data about a certain subject, and so on. Thus, whilst the solutions which AI researchers seek are couched in the vocabulary of computer programming, the problems they pose themselves are often couched in the vocabulary of representation, in which the states of the system are characterized in terms of what they are supposed to stand for.