Tanimoto presents an overview of artificial intelligence, with examples of AI programs written in Common Lisp. Like most books on the subject, this work covers certain areas more heavily than others, and leaves some areas virtually untouched.

The introduction presents some general points about AI as a discipline, including a brief mention of its philosophical underpinnings. Chapter 2 gives a quick review of Common Lisp, one of the main programming languages used for AI work. The following chapters describe the main staples of the field: productions, logical representation and logical reasoning, and search. Chapter 7 is a welcome deviation from most books on AI, as it describes probabilistic reasoning at a reasonable level of detail and puts it in the perspective of handling uncertain knowledge. Chapters 8 through 11 cover other traditional areas of AI, such as learning, natural language understanding, vision, and expert systems.

My reaction to the book is mixed. Its coverage of the field is similar to that of other books, and it is traditional for each author to emphasize his or her areas of expertise. Certain parts of the book appear a little dated, probably because it was published in 1990. My main concern, however, is that some of the material is presented at a relatively superficial level. The important information is not missing, but it is usually buried amid less important details. In several places, I would have preferred a little less breadth and more depth.

Take, for example, the chapter on Common Lisp programming. This chapter is intended to provide an introduction to the language, rather than a full description (which would require a large volume by itself). The point of the chapter is to enable readers to understand the many algorithms and programs that are presented throughout the book, which are written in Common Lisp. This approach is flawed, however. The material is clearly insufficient to enable readers who do not already know Lisp to understand the fine points of the examples and algorithms. Unfortunately, the explanations that accompany the algorithms are often insufficient, by themselves, to tell the reader exactly what is happening. I would have preferred a different approach; for example, the book could have relied on pseudocode (which does not require a chapter to explain) for all the examples, and given more detailed English explanations. Alternatively, more space could have been dedicated to LISP, and other parts of the book could have been reduced correspondingly.

In spite of these criticisms, the author does a reasonable job of presenting much of the material. One of the book’s strengths is the number and variety of exercises at the end of each chapter, ranging from pencil-and-paper exercises to project-sized problems. When used in a course (the author suggests one or perhaps two semesters) with emphasis on the exercises and with supporting material, this would make a good textbook. Each chapter contains good lists of references and a brief review of the most important publications on the subject. The writing style is simple and to the point, and typesetting and index are good.

In conclusion, this is neither an exceptionally bad nor an exceptionally good book on artificial intelligence. It could benefit from a stronger attempt to unify the key ideas in the field, and from a differential treatment that emphasized certain points more than others. While I would not recommend it by itself to someone who wants to learn all about AI, in conjunction with other material (perhaps another textbook) it would present a good introduction to the field.