Recursive functions of symbolic expressions and their computation by machine, Part I
Communications of the ACM
Computers and Thought
Minds and Machines
| Hi-index | 0.00 |
By “intelligence” we mean a property of a system, a judgment based on observation of the system's behavior and agreed to by “most reasonable men” as intelligence. “Artificial intelligence” is then that property as observed in non-living systems. Work directed toward producing such behavior is thereby work in artificial intelligence. While the above is indeed a loose definition, more useful in suggestiveness than in precision, it should suffice for our purposes. It does contain at least one strong assumption—that, a priori, intelligence is not restricted to “living” systems. And it does suggest that, if the issue of whether artificial intelligence does or will ever (or can ever) exist is really worthy of further argument or concern, then some agreement should be reached concerning the rules for observation and agreement among most reasonable men. Here, however, we shall treat the term as merely a convenient label. This paper is intended as an overview of artificial intelligence, in the summer of 1966, by one peripheral to the field. It summarizes the author's reaction to a review of a part of the printed record. As such, it cannot be exhaustive. (In the Permuted and Subject Index to Computing Reviews 1964-19654 alone, there are over 275 entries under artificial intelligence.*) Nor is it necessary to provide another extensive bibliography of the field. Published material by Minsky, 15 Feigenbaum, 7 Newell,17,18 and others, together with Computing Reviews3,4 serve this purpose quite well, This paper is rather a set of personal comments, and, as such, an invitation to discussion, rebuttal, and controversy.