Abstract

Since the introduction of the computer in the early 1950's, the investigation of artificial intelligence has followed three chief avenues: the discovery of self-organizing systems; the building of working models of human behavior, incorporating specific psychological theories; and the building of "heuristic" machines, without bias in favor of humanoid characteristics. While this work has used philosophical logic and its results may illustrate philosophical problems, the artificial intelligence program is by now an intricate, organized specialty. This book, therefore, has a quite specialized audience of its own although it can be very valuable to those philosophers who are interested and competent in using this pioneering material. Five scientific papers report attempts to solve five different kinds of problems. Bertram Raphael describes an attempt to build a memory structure that converts the information input into a systematic model by "understanding" the informational statements as they are made. Daniel Bobrow's machine can set up algebraic equations from informal verbal statements. M. Ross Quillan asks: "What sort of representational format can permit the 'meanings' of words to be stored?" Thomas Evans' machine, Analogy, serves as a model for "pattern-recognition" rather than the "common-property" method of semantic memory. Fischer Black has developed a logical deduction mechanism for question-answering which keeps track of where we are and avoids endless deduction. The editor and John McCarthy contribute more general chapters, providing the historical background of cybernetics, and dealing with the problem of formalizing a concept of causality. Minsky ends the volume with his view that our convictions on dualism, consciousness, free will, and the like are used in the attempt to explain the complicated interactions between parts of our model of ourselves.--M. B. M.