Mind Matters: Exploring the World of Artificial Intelligence

James P. Hogan

Ballantine Publishing Group

New York

1997

Reviewed By: Magan Stephens

August 1998

Mind Matters by James P. Hogan looks deeply in to history of Artificial Intelligence (AI) from the history of logic and philosophy to the current problems that face AI today. Hogan’s objective is to give his readers the basic background in the concepts and processes that are involved in AI research with regard to the linguistic, philosophical, physical and computing limits that have recently been realized. He also takes a look at expert systems, parallel computing, genetic programming, and holographic processing. These AI procedures are just a few of many approaches to branch out of Artificial Intelligence research.

Expert systems are computer programs that simulate the intelligence and behavior of a human and contain expert knowledge in a particular field. They usually are comprised of a knowledge base containing a set of rules for applying the knowledge base to particular situations that is described to the program. Popular expert systems are those that play chess such as Deep Blue and those that assist in medical diagnosis for instance Stanford’s MYCIN. MYCIN’s purpose is to advise physicians on the diagnosis to brain and spinal cord infections and to provide a list of antibiotics for treatment.

Parallel computing consists of multiple processors that break down applications into many separate and independent operations of massive amounts of data. One of AI’s biggest problems is having enough speed and memory to accommodate bigger programs and large mountains of data. A computer like Big Blue, for instance, requires parallel processing to analyze multiple moves in a chess game (308). Roger Schank, an AI researcher from Yale, is an opponent to the parallel computing theory. He feels that the optimism over parallel computing is misdirected. Increased speed does not represent the obstacles that AI is up against such as world knowledge, learning, or commonsense.

Holographic processing and genetic programming are more recent approaches to Artificial Intelligence. Holographic processing is similar to human vision. Holograms are three-dimensional images and regenerate light waves and may solve AI vision recognition problems since it exhibits associative abilities like the human mind. Genetic programming shows signs of an evolving computer or "programs that will write themselves" (346). Genetic programming mimics natural selection to evolve strategies for solving complicated

problems. These are ambitious notions but Hogan includes them in his book to show how far AI has come.

Rather than taking a "strong" or "weak" position of AI, Hogan focuses on the many brilliant minds behind AI endeavors. However, his optimism towards building a machine that thinks like a human is evident throughout the book. He posits that humans "come into the world knowing nothing and assessing only the ‘ground rules’ that were put their by God or by nature, depending one’s beliefs. And most of what they learn after that comes through language, books, teachers and example" (140) implying that machines can "learn" through data input just as humans do. Hogan also focuses on the failures to predict the future of technology. This failing has only encouraged more contention among the AI community.

Hogan’s hopefulness of AI research seems to mirror that of today’s progressive thinkers such as Doug Lenat and Vernor Vinge. Vinge ‘s "Singularity" theory states that technology is advancing at such an accelerating curve that the machines that we have so diligently tried to replicate in our own image will eventually build better machines, thus bringing about the end of the human era (351). Although "Singularity" sounds a bit extreme, Vinge foresees this phenomenon as inevitable if we can, in fact, build a superhuman machine. Lenat, once an AI pessimist, discovered in his CYC project that stringing together semantic concepts and forming associations will lead to the development of language comprehension in the same way a human infant acquires it. This evolutionary milestone of grammar comprehension will only accelerate as new information is supplied and CYC learns to read (256). Conversely, Hogan enumerates the pessimism, which shrouds a contingent of the AI community.

The fact that intelligent machines are possible raises many concerns. People feel threatened when their unique sensibilities are called into question. Hubert Dreyfess, a student of physics and later a Philosophy professor at MIT holds the position that humanlike intelligence cannot be imitated through symbolic representation. He feels that people have "no access to the basic elements and first principles underlying their perceptions, actions, and use of everyday knowledge" (290). This tenet was in direct opposition of the rule-based efforts of Noam Chomsky’s Transformational Grammar and David Marr’s Theory of Vision, both of whom were at MIT the same time as Dreyfess. AI has been and always will be a controversial subject.

This non-fiction book is for the technological handicapped as well as technologically adept who are interested in Artificial Intelligence from its inception. Hogan describes concepts in an easy and sometimes humorous manner. He often illustrates hard to grasp concepts in a simple fashion, such as comparing a Parallel Distributed Processing (PDP) to the composition of the human brain. His chapter "Logical Mind", provides a rudimentary course on beginning logic for those who have not experienced syllogisms or Venn diagrams before. I recommend this book to everyone who has faith in technology and a proclivity towards the ambitious endeavors of mankind. Hogan succeeds in inspiring his audience to think about the implications of Artificial Intelligence and therefore enabling them to draw their own conclusions.