After Thought: The Computer Challenge to Human Intelligence (review)

After Thought presents provocative conjectures about emerging computer methods that James Bailey believes will unravel the mysteries of physical and biological science, as well as those of economics and other social sciences. He has a felicitous writing style, and Basic Books has produced a handsome, well-designed volume. However, this treatise is seriously flawed. Instead of simply explaining and extolling several successes of neural networks, classifiers, self-correcting programs, and other innovations made possible through powerful parallel processing machines, Bailey asserts that nothing less than “bit evolution” is leading to a “new brain form.”

Bailey discusses aspects of the history of science and the history of computer technology and asserts that these areas of human endeavor are impeded through ties to linear, sequential thinking. We are implored to free ourselves from these linear and sequential modes. If we do, he assures us that based upon analysis of current computer capabilities bit evolution will carry civilization into a new glorious era of “intermaths,” self-propelled computer methods. Part of the new world that Bailey describes is one in which computers learn from raw data without human intervention.

While After Thought does make a point, Bailey does not develop it convincingly. All his physics examples are taken from basic mechanics, while his astronomy citations are limited to comments about Kepler, Galileo, and Newton. He makes no reference to modern physics and the ability of mathematics and computers to deal with many-body theory, quantum mechanics, elementary particle physics, or myriad topics in solid-state physics. Similarly, he makes no mention of formulations about modern cosmology. His vision of pre-intermaths mathematics is limited to algebra, geometry, and calculus. His view of mathematical methods capable of dealing with complex phenomena leaves out simultaneous differential equations, the applications of multidimensional vector spaces and statistics, and the use of factor analysis and principal component analysis to tease out significant causal agents in complex interactive systems. Techniques of fuzzy logic and linear programming methods are also ignored. Nowhere does he discuss the role of feedback in creating models of complex systems that evolve in time. From this extremely limited perspective on science and mathematics, Bailey concludes that intermaths will be the inevitable basis for understanding tomorrow’s world.

Bailey swings back and forth between issues in hard sciences and those in the social sciences of economics and war/peace studies. It is not clear what he really anticipates, and there is no convincing basis for agreeing with his formulations. He extols the power of neural networks but ignores the fact that we know very little about how the mind functions. New paradigms for computing can be quite helpful when one is seeking a pattern from a set of confusing and sometimes incomplete data. However, success in predicting stock market behavior or in identifying credit card fraud is scant proof that computers “will evolve a process of intelligence that is not the same as ours, or even understandable by ours” (p. 10).

What is distressing about this book is not just its theoretical exposition but its call for new types of education that will free children from the constraints of sequential thinking and open up great opportunities with the intermaths. Implicit in Bailey’s discussion of education are erroneous and misleading premises about the psychology of learning and the use of the Internet in education. For example, his discussion of the advantages of NASA’s Live from the Stratosphere Internet project and criticism of the Northwestern University Internet-based CoVis program is ill informed and inappropriate. He calls for open-ended problem solving and attention to complex phenomena, but he does not seem to understand what is actually accomplished along those lines in the Stratosphere or CoVis projects. Nor does he mention STELLA, the systems thinking software based upon the modeling methods of Jay Forrester. STELLA is engaging many schoolchildren in exactly the type of education that Bailey thinks will occur only through applications of intermaths. This...