Can you reason about the real physical world on the classic logic of true and false? Perhaps the two-valued evaluation characterized by artificial intelligence (AI) systems is inadequate for the real world “because of the aspect of uncertainty. There are two sources of this problem: imperfection of knowledge about the real world, which is gained by [AI systems], and vagueness of [the] notions used for describing objects/phenomena of the real world.”
How do you solve the problem of imperfect knowledge--whether the imperfection is due to uncertainty of knowledge, imprecision of knowledge, or incompleteness of knowledge? And how do you deal with the two basic aspects of the problems of vagueness of notions in knowledge-based systems, namely unambiguity and degree of precision (detail, accuracy) of the process of notion formulation?
The above issues--of crucial importance for understanding the various aspects and issues of AI development--are brought out concisely by the author of this textbook, which is useful for an introductory semester course on AI for undergraduate students from computer science and related fields.
Spread over 17 short chapters in three parts, including ten appendices, a bibliography, a subject index, and a short preface, the author has included the history of AI (and its complementary discipline of computational intelligence), key AI methods, and application areas of AI systems. Interestingly, the author attempts to cover the subjects of epistemology and psychology, too, while presenting the theories of intelligence in philosophy and psychology along with related concepts such as mind, cognition, knowledge, and so on.
As a noteworthy feature, in the last chapter, the author stresses the AI issues, potential barriers and challenges, as well as the schema of various determinants (disciplines) of AI development. His mention of the contributions of various disciplines like linguistics, neuroscience, biology, mathematics, physics, computer science, logics, psychology, and philosophy to AI is quite useful for next-age AI researchers to pick up research ideas from multi-disciplinary domains.
The book is very useful for computer science, engineering, and allied science students, particularly because of the extensive appendices giving formal mathematical models for AI methods. However, because AI is a multi-disciplinary area today, students and researchers from disciplines like psychology, philosophy, and linguistics who are interested in studying and researching AI further will also benefit from using this book for a first AI course, though the appendices may not appeal to them to the same extent.
For new researchers, the bibliography given at the end of the book is quite extensive, covering 324 publications from 1916 to 2010. However, I wonder at the absence of any publications from the last six years. Has the momentum of AI development slowed down? Perhaps the bibliography needs an update in the next edition.