MACHINERY OF THE MIND: Inside the New Science of Artificial Intelligence by George Johnson

MACHINERY OF THE MIND: Inside the New Science of Artificial Intelligence

Email this review


The message is clear from Johnson's sophisticated history/survey that artificial intelligence (AI) is a growth industry. Johnson, who spent a year on a journalist fellowship on research and interviews, turns out to be a first-rate guide through it. His early chapters take the reader through the rudiments of computer hard-and soft-ware, rehearsing some of the history and the fabled names. A signal event occurred in 1956 when, with a small grant, John McCarthy, Marvin Minsky, Allen Newall and Herbert Simon helped organize a conference at Dartmouth that coined the term"" artificial intelligence. ""Each in turn went on to spread the gospel and gain disciples at M.I. T., Stanford and Carnegie Mellon. McCarthy himself developed ""Lisp,"" a powerful, flexible language widely adopted by AI programmers. Successive chapters deal with the major challenges of AI: how to get computers to recognize normal language, translate, build memory, learn by doing, perceive, use logic, generate theorems, art, music. . . The scope is enormous but Johnson provides rich and readable detail. Inevitably he covers some of the older familiar ""classics"" such as Eliza, the psychoanalyst-mimicking program, or SHRLDU, the program that could recognize blocks and pyramids. But, gratifyingly, there are a lot of new programs such as Douglas Lenat's EURISKO (eureka plus heuristics) and AM--automated mathematician--in which he programmed a computer with the principles of set theory and rules of thumb that led the computer to discover the natural numbers and rules of arithmetic. Johnson describes the various guiding philosophies of the principal schools of AI--which ones run by logic, which are more intuitive; which have plumbed psychological processes; which are using the computer to model the brain, and which the reverse. Moral/ethical issues also come up because Johnson is not unmindful that major funding for AI work comes from the Pentagon. International rivalry is also discussed, as is the issue of how to deal with computers if and when they get as smart as people. All of this makes for a worthy volume. It belongs on the shelf with such classics as Margaret Boden's Artificial Intelligence and Natural Man.

Pub Date: Sept. 26th, 1986
Publisher: Times Books