cover

Comments on AI: A Modern Approach

This page contains comments by readers of Artificial Intelligence: A Modern Approach. The comments are organized into sections based on who is making them: instructors, students, reviewers, authors of other AI texts, and others, as well as comments on particular chapters of the book, and the author's comments in an interview by Amazon.com.

If you have a comment to add please mail it to aima-talk@yahoogroups.com (you can see the archives of this discussion group) or submit a review to Amazon.com. or to Barnes & Noble.

You can also see a comprehensive list of reviews at Google Books.

Comments by Instructors

This monumental work, which completely dominates the AI textbook market, has been compared with classics like Watson's Molecular Biology of the Cell. - Prof. Manny Rayner (Univ. of Geneva)

Amazing job!! [The new edition] makes it the AI text for at least the next one or two decades. It's good to see the history of excellence continued. - Prof. Bart Selman (Cornell)

Once again [Russell & Norvig] managed to make an excellent textbook still somewhat more excellent. ... I have the highest respect for the genius way you manage to make complex issues (like boosting) simple through intuitively clear descriptions, and do so over such a broad range of topics. I am convinced that the current revival of interest in AI also has much to do with the availability of such an outstanding textbook for our field. - Prof. Wolfgang Bibel (Darmstadt)

If you only own one book on AI, this is the one you should have. It is extensive, thorough, and full of interesting and useful insights. - Prof. Avi Pfeffer (Harvard)

The publication of this textbook was a major step forward, not only for the teaching of AI, but for the unified view of the field that this book introduces. Even for experts in the field, there are important insights in almost every chapter. I recommend it to anyone who wants to have an introductory overview of the state of AI. And I recommend it to experts in the field, who will enjoy its unified description of the field. I especially enjoyed the introductory chapter and the chapter on philosophical issues. I have taught from this book three times now, and it has improved my AI class hugely. - Prof. Thomas G. Dietterich (Oregon State), in Amazon.com customer reviews

It's a great book, with incredible breadth and depth, and very well-written. Everyone I know who has used it in their class has loved it. I think there's a good chance the book will take over the AI textbook market. - Prof. Haym Hirsh (Rutgers)

It's a pleasure teaching from your book. - Prof. Barbara Grosz (Harvard)

a damn good book - Prof. Pat Hayes (Western Florida)

It's simply the best. - Prof. Curry Guinn (Duke)

The book is the most comprehensive and most insightful introduction to artificial intelligence that I have seen. It provides a unified view of the field organized around the rational decision making paradigm. It covers the traditional topics of search, logic, planning, and knowledge representation along with current research in reasoning under uncertainty, machine learning, robotics, and more. - Prof. Elisha Sacks (Purdue)

Russell and Norvig's book is terrific: well-written and well-organized, with comprehensive coverage of the material that every AI student should know. It includes pseudo-code versions of all the major AI algorithms, presented in a clear, uniform fashion. The authors have done an excellent job of relating work in AI to work in other fields, both in and out of computer science. It's a pleasure to teach from this book. - Prof. Martha Pollack (Michigan)

A remarkably comprehensive and incisive treatment of the field. By organizing the material around the task of building intelligent agents, Russell and Norvig present AI as a body of inter-related design principles, rather than a loose grab bag of techniques and tricks. Students hungry for meaty ideas will find ample nourishment from this text. ... A masterful pedagogic achievement. - Prof. Mike Wellman (Michigan)

I'm delighted to report that Russell and Norvig's book is easily the most comprehensive textbook on Artificial Intelligence that I've seen. The material is well-organized, and, if the entire textbook cannot be covered it is easy to pick and choose selected topics to teach a short or introductory course. Having taught from an early version of the textbook, I was very pleased with the in-depth coverage. The presentation is clear and very well-written, and my students had very positive comments. Moreover, I found that I learned a lot myself! - Dr. Steve Minton (USC)

It appears to be the best book on the market in terms of breadth and depth of coverage. I like the focus on the topics that are mostly likely to have an impact on building and analyzing current and future AI systems. I am also a big fan of covering ideas and techniques that can be stated precisely and whose capabilities and limitations are relatively well understood (especially in intro books). Unlike many other books you make an effort to discuss the relationship of AI problems to other very similar problems tackled by other fields. E.g., mentioning learning in the context of function approximation, relating TD learning to bandit problems, etc. - Prof. Simon Kasif (Johns Hopkins)

My reaction to the first edition of Russell and Norvig was, "This is by far the best book on the market, but it would not be impossible to write a better one." My reaction to the second edition was, "No one could write a better AI textbook," and the careful reading I have given it for this review has only confirmed that judgement. This is a great textbook, of amazing depth and breadth, in a league with Feynman's Lectures on Physics. It is not just a textbook for AI classes; it is an unparalleled survey of the theories of rational thought and action. - Prof. Ernie Davis (NYU)

Russell and Norvig [present] this rapidly progressing science in a clear, inviting style. The book combines solid theoretical analysis with practical examples that together tackle and illuminate the core issues. It brims with confidence, optimism and contagious excitement about the frontiers of AI without compromising any of the complexity and depth of the field. This book will be welcomed and enjoyed by students and professors alike. - Prof. Shlomo Zilberstein (Massachusetts)

I used [AIMA] this semester in my Intro AI course and you should be happy to know the students loved it. ... Its A Real "Rave". I've been looking for a better text from the prior ones and [AIMA] did the trick. ... In our PHD program [at Columbia] we require the students to pass 4 qualifying exams (written) in the areas of Software Systems, Computer Hardware, Theory and AI. Each qualifier area covers a very broad range and is accompanied by a substantial reading syllabus. I proposed, and won acceptance by the AI qual committee, to replace a number of older materials on our syllabus with a major portion of [AIMA]. This served to shave nearly 1000 pages from the overall syllabus, still managing to cover the same broad areas with a few new ones. - Prof. Sal Stolfo (Columbia)

I used AIMA for my fourth-year AI class last term with great success; the book was comprehensive, well-laid out, and above all interesting. - Prof. John Anderson (Manitoba)

I like this book very much. When in doubt I look there, and usually find what I am looking for, or I find references on where to go to study the problem more in depth. I like that it tries to show how various topics are interrelated, and to give general architectures for general problems ... It is a jump in quality with respect to the AI books that were previously available. - Prof. Giorgio Ingargiola (Temple)

It is an impressive book, which begins just the way I want to teach, with a discussion of agents, and ties all the topics together in a beautiful way. - Prof. George Bekey (USC)

Really excellent on the whole and it makes teaching AI a lot easier. - Prof. Ram Nevatia (USC)

A marvelous achievement, a truly beautiful book! - Prof. Selmer Bringsjord (RPI)

I give a rave review to [AIMA], which I used in my grad/undergrad course last semester. More importantly, the students liked it too! - Prof. William Rapaport (Buffalo)

[AIMA] is fantastic! I'm enjoying the book immensely. It is a real contribution to the field. - Prof. Tom Dietterich (Oregon State)

My current favorite as an AI textbook. - Prof. Dana Nau (Maryland)

Just terrific. The book I've always been waiting for. - Prof. Gerd Brewka (Leipzig)

A first-rate job. - Prof. Paul Kube (UCSD)

[An] outstanding textbook! I found it both comprehensive and up-to-date. - Prof. Oren Eztioni (Washington)

An impressive piece of work. - Prof. Jim Martin (Colorado)

Materials are presented in a very coherent manner. It surely sets a new standard for textbooks in CS! - Prof. Sung Myaeng (Chungnam National, Korea)

I am using your excellent book and the students are as delighted as I am. - Prof. Christoph Herrman (Darmstadt)

Very informative, extremely well-written and very well organized - Prof. Prasad Tadepalli (Oregon State)

A very good book which I will use again. BTW, my class is an introduction to AI, with upper level undergrads (few) and beginning grads (lots, MS level) ... only a couple are interested in AI research. - Prof. Michael Gray (American)

Student Comments

Not just a good AI book, but I would consider this one of the most interesting and thoughtfully structured textbooks I've seen on any subject. - A reader from Chicago, commenting on Amazon.com The most useful book I own. - Marek Peetrik, on Amazon

Among the best textbooks I've ever used. - Neil Conway, on Amazon

This is a great book for anyone who wants to get serious on AI algorithms - Rodrigo Damazio, on Amazon

I have to say that the book was the best 80 dollars I have spent on a textbook so far. So good, in fact, that I have been reading it through the summer. - Student (UMass Dartmouth, in comp.ai)

Not only one of the best on AI, but one of the best computer related books I have ever read. - Student (Univ. Michigan, Flint)

The best text book I've seen on this, and any other, subject. - Student (St. Andrews)

Outstanding. I couldn't be happier with the clear, thorough, well-organized treatment of the subject. I knew virtually nothing about many of the subjects in this book before this course, and this book has helped pique my interest sufficiently that I hope to go into the field. The references to important works in the field were extremely useful, especially since many of them are now available online. As a reference, the only downside is that the pseudocode is often not detailed enough to make the implementation clear, although the availability of the Lisp code online offsets this in many cases. - Student (via Internet)

I have truly enjoyed ... "AI A Modern Approach". I am a 2nd year computer science graduate student at UCLA, and since learning of this masterpiece last year, I have referenced it on almost a weekly basis. - Student (UCLA)

The book is incredible! - Student (SUNY Buffalo)

I love your book and find it better than all the other books in AI. - Student (South Florida)

Your book is excellent. It's very well written and reads better than any other book I've seen on the subject. - Student (Berkeley)

It is a nice change from the normal boring textbooks out there! - Student (Pennsylvania)

Awesome book. It wasn't required for the class I'm taking, but I bought it anyway. It's by far the most complete (and enjoyable) book I've ever seen on AI. - Student

These exercises are complete and thorough. They cover the entire breadth of the chapter, and each are detailed enough to challenge the students as more than a trivial problem. - Student (Pennsylvania)

It presented useful information about concrete aspects of AI while providing very interesting historical background. Unlike most textbooks I've seen in CS, this one seemed to aim at actually relaying information. ... This book was a refreshing change. - Student (Columbia)

The way the chapters relate to each other and the theme of intelligent agents develops from one chapter to another makes the book flow well and makes the entire field seem more organized than I had ever imagined. ... excellent. - Student (Columbia)

I really liked it. Its very informative, but still quite readable. Definitely one of the best textbooks I've had, period. I especially like the history chapter, although it's not really applicable to the course. Worth the money. :) - Student (Columbia)

This book offers the most coherent view on AI, as well as the broadest range of state-of-the-art algorithms. - Student (Oslo)

I am just mailing to congratulate you on a really useful book! I used it in my undergraduate studies and am currently doing post-grad. I have again today turned to your book for explanation of some concepts that I can't find a better explanation for anywhere on the internet! I think people, especially in the scientific and engineering society, underestimate the importance of simple explanations of difficult concepts, especially with regard to people that are new in the field. Thanks again for making difficult concepts seem not so difficult! - Student (South Africa)

Comments by Reviewers

We provide here excerpts from reviews that have been published in Knowledge Engineering Review, Computing Reviews, Cybersecurity Canon, comp.ai, comp.ai.games, rec.arts.books.reviews, AI Magazine, AI Expert, and Artificial Intelligence.

You can also see all reviews in the Internet news groups.


Yet another introduction to Artificial Intelligence? Don't we have enough of these already? This was what I thought - before I held a copy of the book in my hands for the first time.

In fact, this book is different in many aspects from every other general AI book you may have seen before. First of all, it's unique in the broad coverage of topics. It (almost, see below) has it all: the book's 27 chapters cover problem solving and search, logic and inference, planning, probabilistic reasoning and decision making, learning, communication, perception and robotics. And in each section you will find an incredible amount of useful and totally up to date material that has never been included in other textbooks so far. There is a lot to learn, for beginners and for advanced Al people. You always wanted to know about Socratic reasoners, demotion, the upward solution property, coercion, policy iteration, PAC learning, adaptive belief networks, convolution, bigram models, the Viterbi algorithm, skeletonization, the horizon problem and the like? Then this is the right book for you.

Also the more standard parts have a lot of "nonstandard" material. The logic sections, for instance, not only give the typical introduction to propositional and first order logic together with the usual inference procedures, they also give many useful hints how to use first order logic to actually represent aspects of the real world including measures, time, actions, mental objects and the like, and they contain a lot of information about how to implement efficient logical reasoners. The section on uncertain knowledge contains an excellent introduction to probabilistic reasoning and belief networks. Moreover, it introduces decision theory covering topics like multiattribute utility functions, decision networks, sequential decision problems and dynamic decision networks. The section on learning, one of my favourites, presents all sorts of approaches ranging from "subsymbolic" back-propagation learning in neural nets, genetic algorithms, decision tree learning, explanation based learning to inductive logic programming, and it puts all these approaches into perspective.

The book also contains very valuable information about how specific approaches and techniques were used in real applications, how successful they were - or why they failed. The planning section is a particularly good example of this. With this information the reader gets a pretty good feeling of what can be done at the moment, and where the big problems are.

A further important aspect distinguishing this book from others is the common unifying perspective under which all the different approaches are presented. The authors view Al as the science of intelligent agent design. Under this view all the bits and pieces from various subfields of AI fit together very nicely. For novices this provides a lot of orientation. Advanced researchers get the great feeling that what they do is not only relevant to them and their little Al subcommunity.

The book is very well-written and clear with an excellent balance between motivation, formalization and application. To make the underlying ideas precise the authors use easily understandable pseudo code throughout the book. Actual Common Lisp implementations of the presented algorithms are available via the Internet. The authors show a great ability to invent illustrating and entertaining examples - often reappearing in several chapters - and their style of writing is very amusing. It's just fun to read this book.

After all this enthusiasm for the book, is there any wish left open? There is. I don't want to mention the few mistakes one finds, rather unsurprisingly given this is the first edition of a book of over 900 pages (of course, consistency of first order logic is NOT semi-decidable, contrary to what is stated on p277). Number 1 on my wish list is a more adequate treatment of nonmonotonic reasoning. Much excellent work has been done in this area in recent years and interesting insights have been gained. All Russell and Norvig, basically, have to say about this is: theoretically interesting but practically irrelevant. I think this is too much of an oversimplification and the topic of nonnumerical defeasible reasoning deserves more than one page in a book like this. I would hope to see one or two extra chapters on this topic in a future edition.

Anyway, there can be no doubt that this is the best general AI textbook available today. If the quality of textbooks mirrors the matureness of a field AI is in much better shape than many of us may have thought. The book is a highly valuable source of information, not just for newcomers. Given its reasonable (if not cheap) price there is a pretty good chance that this will become the AI bible of the next decade.

- Gerd Brewka (Technische Universitat Wien, Austria) in The Knowledge Engineering Review, Vol. II: 1, 1996, 73-83


The authors have not only written an excellent textbook, but have distinguished themselves by their appraoch to presenting all the major themes of AI.

- P. Navrat, Bratislava, Slovakia, in Computing Reviews, May 1997


Executive Summary

For many, Artificial Intelligence: A Modern Approach is the de facto bible of artificial intelligence. It combines in-depth treatments of introductory and advanced concepts, along with historical background and accessible explanations. Including algorithms, code and pseudo-code, the book sits between master's and Ph.D. level, but is accessible to all. Your journey on the road to the application of data science should start here.

Review

Artificial intelligence and machine learning technology now permeate our lives. We are increasingly using them, or subject to them, whether we realise it or not. Increasingly ubiquitous implementations include practical speech recognition, machine translation, self-driving vehicles and household robotics. Artificial Intelligence: A Modern Approach helps provide clear understanding of exactly what AI and machine learning comprise, and what they can and cannot achieve. Such clarity of thought helps us move from buzzword-dropping to actual scientific understanding. Underlying concepts are explained with clear analogies and accessible language.

From algorithmic and coding perspectives, the tools provided are powerful, though we remain some distance from machine sentience, which should never be confused with AI. Imitation remains an intriguing concept, although it is increasingly unclear whether it tests machine or human intelligence. Russell and Norvig's book will help you gain insight about this field and enable you to apply your own critique and assessment to Turing's test.

Algorithmic research has also seen numerous key developments since 1950, not the least of which are game theory-based thinking (particularly the work of mathematician John Nash) and the solution of the game of draughts. Much theoretical progress has also been made, particularly in such areas as probabilistic reasoning, machine learning and computer vision. The book will help you develop an appreciation for the critical role of data modelling over algorithm selection, and where the real value lies in machine learning.

The book is as close to exhaustive as is currently available in the field, including in-depth treatments of non-technical learning material whilst providing an accessible and understandable overview of major concepts.

Since the 2003 edition, increased coverage has been given to topics such as constraint satisfaction, local search planning methods, multi-agent systems, game theory, statistical natural language processing and uncertain reasoning over time. Attention has also been given to providing more detailed descriptions of algorithms for probabilistic inference, fast propositional inference, probabilistic learning approaches including EM, and other topics.

The book contains up-to-date and extensive exercises, delivering a unified, agent-based approach to AI: organising the material around the task of building intelligent agents. The comprehensive, up-to-date coverage includes a unified view of the field organized around the rational decision-making paradigm.

The authors' approach delivers in-depth coverage of basic and advanced topics, and provides a basic understanding of the frontiers of AI without compromising complexity or depth. It conveys in-depth understanding and clear explanation of such concepts as supervised and unsupervised machine learning, and thus to the layman, an understanding of why there will be no jobs for machine learning foremen!

Pseudo-code versions of the major AI algorithms are presented in a uniform fashion, and Actual Common Lisp and Python implementations of the presented algorithms are available online, as are test data sets and samples.

Although the field of research has grown considerably since its launch in Turing's seminal 1950 paper, this volume represents both an access point for all interested and in-depth information for those with considerable exposure to the topic. It provides a lens which can be viewed from two directions: 1) towards the past and the history of the field to understand how we have come to where we are today, and 2) towards the future to better understand what is currently possible, and where research is taking us going forward.

This highly popular text, both at undergraduate and post-graduate level, does not claim to be all-encompassing or exhaustive. However, it is a comprehensive treatment given the wide range of the topic. It comes as close as possible, at this time, to being a one-stop reference. As Einstein famously said, "Everything should be made as simple as possible, but no simpler." In the same vein, the book conveys how we can strive towards as much automation as possible, but no more than is necessary. Tacit knowledge and domain expertise remain, for the foreseeable future, beyond the grasp of AI. When it comes to context and corroboration, the input of the human analyst is invaluable. The discipline of data science requires both human and machine input. Completion of this text will help you appreciate why.

Conclusion

Highly recommended. Intellectually, Artificial Intelligence: A Modern Approach provides both a conceptual artificial intelligence gym and a running track to limber up on. The more you use it, the more you will get from it.

- Adrian Culley (Palo Alto Networks) in The Cybersecurity Canon, 2017


Article: 8869 of comp.ai
Newsgroups: comp.ai,comp.ai.edu
From: devika@cs.cornell.edu (Devika Subramanian)
Subject: A review of Russell and Norvig's new AI text
Organization: Cornell Univ. CS Dept, Ithaca NY 14853
Date: Wed, 30 Nov 1994 15:14:29 GMT
A brief review of

Artificial Intelligence: A Modern Approach
Stuart Russell and Peter Norvig
Prentice Hall, December 1994. ISBN 0-13-103805-2

by Devika Subramanian, Cornell University

While the enterprise of artificial intelligence has often been defined around the dream of intelligent agents, Russell and Norvig's book is the first attempt to present the technical accomplishments of AI to a broad scientific audience in the context of embedded agents acting in real-world environments. The book is not merely an expositional triumph; Russell and Norvig achieve a unique synthesis of concepts and algorithms in AI that have evolved in very disparate sub-communities of the field. The book draws on ideas from logic, decision theory, control theory, Markov processes, economics, on-line algorithms, complexity theory, probability and statistics and information theory, to coherently present methods in AI in a jargon-free manner. This makes the book an ideal introduction to newcomers to AI from computer science as well as other branches of science and engineering. For seasoned practitioners, it offers a new, thought-provoking way to understand AI.

The book is organized into eight sections. The first section begins with a brief history of AI and introduces the basic vocabulary for describing agents embedded in task environments. The last section (Section VIII) comprises a beautiful essay on the philosophical foundations of AI and an engaging commentary on the current state and future challenges facing AI. The sections in between constitute the technical meat of the book. Section II highlights general problem-solving methods for embedded agents and includes informed search methods that take resource constraints into account. The third section emphasizes the role of knowledge in decision-making and presents an array of methods for representing and reasoning with logical or categorical knowledge. Section IV presents planning as reasoning about action choice; contemporary planning and replanning methods are presented as specializations of the general methods of logical reasoning introduced in the third section. Section V introduces probability and decision theory as tools for agents acting under uncertainty. It explains how belief networks can be used to represent uncertain knowledge and describes decision-making methods based on them. The sixth section focuses on learning and adaptation in intelligent agents. It presents a unified model of learning, a brief introduction to computational learning theory, as well as specific techniques such as decision-tree learning, neural networks, and a new method for learning belief networks. It also includes a tutorial exposition of recent work in reinforcement learning, as well as the knowledge-based inductive logic programming method. Section VII focuses on interactions of the agent with the external world: natural language communication, perception and robotics. Russell and Norvig have recruited established experts (Jitendra Malik and John Canny) to cover the specialized topics of perception and robotics, ensuring a uniformly high quality to all of the technical material in the book.

The book is hefty: over 900 pages in all. However, almost 200 pages are devoted to items sometimes missing from AI texts: a very thorough index, a truly massive bibliography, "Historical Notes" sections that are researched in depth and make fascinating reading, and a large collection of excellent exercises.

This is perhaps not the place to go through all the book's chapters in detail, but some deserve special mention. The second chapter on agents is brilliant; it puts the entire history of work in AI in perspective and explains why people built the algorithms that were built. This is the first question that most first-timers to AI have, and this is answered up front. The chapters on reasoning about uncertainty are by far the best tutorial exposition of material on probability and belief networks: they make the original papers in the area much more accessible.

Judged from all respects, this is a remarkably comprehensive and incisive treatment of the field. The book is well-written and well-organized and includes uniform and clear descriptions of all major AI algorithms. The authors have managed to describe key concepts with technical depth and rigour without falling prey to stodginess and Greek-symbolitis. AI is presented as a set of inter-related design principles, rather than a grab bag of tricks. The book brims with optimism and contagious excitement about the frontiers of AI. I recommend it without reservation to anyone interested in the computational study of intelligence, whether they be undergraduate or graduate students or senior scientists in the field.

About the reviewer Subramanian is an Assistant Professor at the Computer Science Department at Cornell University. Her interests are in AI, its theoretical foundations and practical applications in design, scheduling and molecular biology. She has been teaching AI at the undergraduate and graduate levels for about five years.

I believe it's the best AI text now available, and definitely the best for games programming. - Bryan Stout, in comp.ai.games
A terrific book ... remarkably comprehensive ... not only provides sufficient background to begin serious work in AI, but also provides just necessary background: there isn't much in it that could readily be omitted by a graduate student in AI. ... Throughout the book, the writing is clear and engaging, and the authors convey an appropriately positive view of the field. To read this book is to get a sense of the intellectual substance of the field-to realize how much good work has been done in AI. ... My minor complaints aside, I've found it a pleasure to teach from this book, and I have also used it frequently as reference source. ... If you want to teach an AI course around an ``agents" theme-and I don't necessarily think that's a bad idea-this book will make it easy for you to do so, and to do so well. But even if you think that ``agents" is just the latest buzz-word, don't let the fact that this is billed as ``the intelligent agents book" dissuade you from adopting it for your class, or from buying it as a reference book. Artificial Intelligence: A Modern Approach will provide a first-rate education in AI even to the reader who skips all the specially agent-oriented material.

Martha E. Pollack is associate professor of computer science and intelligent systems at the University of Pittsburg. She received the Computers and Thought Award in 1991 and a NSF Young Investigators Award in 1992. Her current research interest include computational methods of rationality, plan generation and recognition, natural language processing, and AI methodology.

- Excerpts from a review by Martha E. Pollack in AI Magazine, Fall 1995 (Full review available there.)
In 900 pages of well laid-out text, with excellent use of typography to make finding topics easy, it seems to be a great compendium of methods loosely called "AI."

- Excerpt from a review by Timothy May (Colorado), in rec.arts.books.reviews


This book may very well be the first of the new breed of modern AI textbooks. It uses as a unifying theme the notion of intelligent agents; an excellent pedagogical starting point that lets the authors develop a very hands-on approach as well as one that naturally lends itself to the modern trend toward distributed intelligent systems.

The coverage of all the basic principles of knowledge-based and learning systems is thorough and includes a wide variety of excellent problems. The logico-deductive approach is treated with exceptional clarity and depth; the text is lighter in its coverage of natural language processing and computer vision. A concluding chapter even touches on some of the deeper philosophical issues in more than a cursory manner.

- Review by Philip Chapnick, in AI Expert Magazine, Jan. 1995


Outstanding ... Its descriptions are extremely clear and readable; its organization is excellent; its examples are motivating; and its coverage is scholarly and thorough! ... The authors (and their helpers) have done a remarkable job, and the field owes them a hearty thanks and "well done!" ... will deservedly dominate the field for some time.

- Excerpts from a review by Prof. Nils Nilsson (Stanford) in Artificial Intelligence Journal


I must say that I am hugely impressed by the text book. It is rare that a textbook makes practising AIers happy. The main reason is the coherence that you brought out by using intelligent agency as the glue that binds the various parts together. I have taught this course three times before, and every time I used to dread the introductory lecture. I was disillusioned with the idea of throwing a bunch of definitions of intelligence and talking about Turing test, and asking them to take on faith the fact that representation and search are some how very important. So, I started using a second lecture that talked about planning as a representative AI problem, talking about the idea of domain independent solutions etc. While that helped the students in finding out why search/reasoning etc. is useful, in my own heart, I knew that this was not a good enough job since (what about NLP? what about learning? why learn about them?) Although I am not using your text this semester, I did decide to use the intelligent agents chapter as the basis for my introductory lecture this time. I am happy to report that for the first time, I felt I did a convincing job. I was able to use your agent architecture discussion as a background for explaining why we need NLP, speech recognition, vision, learning, logic, uncertainty, decision theory and uncertainity. What is more, I think the metaphor is so compelling that in many cases I was able to get the students to venture the correct answers about the role of these apparently disparate things that we are going to be talking about this semester. Yesterday night, I was a happy man!

I started reading various chapters of your book and I am very pleased to note that the agent view is woven integrally through all of them. I think that is a great way of bringing things together, and I hope to relay some of those insights to my students.

While the underlying agent view, and the integration it brings, by themselves make your textbook great, I also found that you have done a great job of explaining traditional techniques. For example, I loved your explanation of the distinction between planning and problem solving in terms of the decomposability of the goal test. Similarly, I thought that your description of hierarchical task network planning is better than that found in even some of the state-of-the-art HTN papers. I really envy your encylopaedic grasp of the subject, but thank you for writing it into a text form.

I think this is definitely a watershed textbook for intro to AI courses. Great job!
- Prof. Subbarao Kambhampati (Arizona State)


Comments by Authors of Other AI Texts

I decided to use your book. ... It's really good. It's going to make teaching the class a breeze. - Prof. Elaine Rich (Texas), author of "Artificial Intelligence"

I used AIMA this Spring, and I think it's the best AI text I've ever used. The agent unifying theme works very well, and the text is simply the best combination I've seen of being comprehensive, up to date, and unified. - Prof. Stuart Shapiro (Buffalo), author of ``The Encyclopedia of Artificial Intelligence''

I particularly recommend Ginsberg's and Russell and Norvig's texts. - Prof. Edward Bender (UCSD), author of ``Mathematical Methods in Artificial Intelligence''

Outstanding ... will deservedly dominate the field for some time. - Prof. Nils Nilsson (Stanford), author of ``Principles of Artificial Intelligence'' and other books

You have a whole bunch of stuff in your book that mine doesn't cover. In fact, you cover a whole bunch of stuff I don't even know about. - Prof. Matt Ginsberg (Oregon), author of ``Essentials of Artificial Intelligence''

The best book available now is Russell and Norvig's "Modern Approach to AI" book. It's almost as good as the book Charniak and I wrote, but more up to date. (Okay, I'll admit it, it may even be better than our book.) - Prof. Drew McDermott (Yale), author of ``Introduction to Artificial Intelligence''

The best 'AI' for Intelligent Agents book out there -Michael Knapik, author of ``Developing Intelligent Agents for Distributed Systems''

Comments by Other Interested Professionals

Q. What is a good textbook on AI?
A. Try Artificial Intelligence by Stuart Russell and Peter Norvig, Prentice Hall. - Prof. John McCarthy (Stanford)

... simply the best general AI book available on this planet. ... I can't live without it. - Sergio Navega, IBM

... by far the best book on AI I know; it's comprehensive, insightful, very well-organized and beautifully written. - Claude-Nicolas Fiechter, Daimler-Benz

By far the best and most complete AI textbook around. Othar Hansson (Chief Bibliographic Officer, Thinkbank)

For a comprehensive and inspiring discussion of intelligent agents (from an AI perspective), take a look at Artificial Intelligence: A Modern Approach. It aims to define the whole AI enterprise as the task of building rational agents (in a broad sense). The text is technically rigourous and up-to-date, treating the notion of rational agency not as a 90s buzzword, but as the central concern of AI (past, present and future). IMHO it is likely to eclipse already existing general AI overviews, thereby making a considerable "agentive" impact on the field. I warmly recommend it to AI- and non-AI-practitioners alike. - Are Sorli (on the agents@sun.com mailing list)

Wow! What a coup! The writing is clear, entertaining, and full of things that were new to me. The overall structure of the book is beautifully coherent. I think it has a real chance to dramatically affect the field by totally reframing what is viewed the baseline for AI. - Dr. Steve Omohundro (International Computer Science Institute)

My first impressions are very positive: the treatment is thorough, with concrete examples, and the writing is superb. - Prof. Ross Quinlan (New South Wales Institute of Technology)

I want to congratulate you on your newest AI text. I think it belongs to the rare family of books that paradoxically help save bookshelf space as one can safely replace a whole mini-library with one book. Second (also very rare) impression: it is an adventure to browse and read this book. This impression begins already with the cover and continues with the uniformity of style, clarity of exposition, as well as clean design of figures. Somehow the book inspires trust in the reader. - Dr. Jacek Ambroziak (Sun Microsystems)

Very impressive!!! The scope and choice of topics is fantastic... but primarily I was impressed with the presentation (of the bits I read), especially the focus on agents and environment, elements of decision-making and so on. It's about time AI texts caught up with the times... great job! - Prof. Craig Boutilier (British Columbia)

I think that the new AI text is excellent, and is probably going to be quite popular-it takes a much better slice of AI than any other text I've seen. - Dr. Greg Provan (Pennsylvania)

I have had a chance to inspect your book and find it very good and very impressive. - Prof. Don Loveland (Duke)

A quantum leap over previous textbooks! - Dr. Peter Karp (SRI)

The best AI text I've seen yet. Excellent work and quite readable. - Prof. Jeffrey Putnam (New Mexico Institute of Mining and Technology)

I think [AIMA] is super. I think it is the best AI book on the market - Prof. Nick Cercone (Regina)

Comments on Particular Chapters

Chapter 1: This chapter is very well researched-going all the way back to Plato. - Prof. Nils Nilsson (Stanford) (in Artificial Intelligence Journal)

Chapter 2: The second chapter on agents is brilliant; it puts the entire history of work in AI in perspective and explains why people built the algorithms that were built. - Prof. Devika Subramanian (Cornell)

Chapter 4: The chapter on search methods made a very well organized and easy to access reference. - Student (Columbia)

Chapter 5: A novel daring new concept in textbook explanations of alpha-beta: making it comprehensible! - Prof. Andrew Moore (CMU)

Chapter 6: They loved it [the Wumpus world agent]! The whole class participated, asking questions, drawing conclusions for the agent, etc. It was very satisfying. - Prof. Bonnie Webber (Pennsylvania)

Chapter 7: ... elegantly show how first-order logic greatly reduces the representational burden. - Prof. Nils Nilsson (Stanford) (in Artificial Intelligence Journal)

Chapter 8: I particularly liked Chapter 8, Building a knowledge base. It unfolded like a dime store detective novel. It was interesting how to digitalize all the different types of data. - Student (Columbia)

Chapter 9: The chapter on resolution and inference rules was very well laid out and made a potentially confusing topic very easy to understand. - Student (Columbia)

Chapter 10: You have captured the ideas of input and linear resolution well on page 285; they are important concepts because of Prolog yet often are omitted or presented incorrectly in basic AI texts, even those dealing with Prolog. - Prof. Don Loveland (Duke)

Chapter 12: [The] description of hierarchical task network planning is better than that found in even some of the state-of-the-art HTN papers. - Prof. Subbarao Kambhampati (Arizona State)

Chapter 14: This has to be the nicest exposition of probability theory I have seen in an AI textbook. - Dr. Keiji Kanazawa (Microsoft)

Chapters 14-17: The chapters on reasoning about uncertainty are by far the best tutorial exposition of material on probability and belief networks: they make the original papers in the area much more accessible. - Prof. Devika Subramanian (Cornell)

The integration of DBNs, HMMS, and Kalman filters is wonderful, and this is the only textbook I know of (in any area, even outside AI) that covers all of this in an introductory and coherent manner. Padharic Smyth (UC Irvine)

Chapter 18: It really is a great book. I was particularly amazed that you were able to convey a coherent description of PAC learning in only a couple of pages. The same subject took 5 weeks to cover in one of our seminars; and it's questionable which one covered it better. - Student (UC Riverside)

Chapter 20: Includes the best general-AI introduction to Reinforcement Learning available today. - Dr. Rich Sutton (GTE Labs)

Chapters 22-23: Russell & Norvig do an outstanding two-chapter job of describing natural language processing. - Prof. Nils Nilsson (Stanford) (in Artificial Intelligence Journal)

Chapter 24: There is no doubt that this chapter does a much more conscientious and intelligent job of treating computer vision than does any other AI book I know. - Prof. Allen Brown (Rochester)

Chapter 26: Although I have not yet read all of your book I am very impressed with what I have seen so far, and glad you decided to include a philosophical chapter. - Prof. Aaron Sloman (Edinburgh)

Algorithms: Russell and Norvig's text fills a huge gap. Not only does it manage a balanced coverage of modern topics, it provides elegant pseudocode for a huge range of important algorithms. - Prof. Daniel Weld (Washington)

Historical notes: These notes are scholarly and exceedingly well researched (by Douglas Edwards). I learned a great deal from them and noted only an occasional minor error. - Prof. Nils Nilsson (Stanford) (in Artificial Intelligence Journal)

I especially appreciate the effort that you put into historical / philosophical perspectives, which are invaluable in a liberal arts curriculum like ours. - Prof. Simon Levy (Washington & Lee)

Bibliography: my first use of [AIMA] has been as a source of key AI papers. I am likely to find the paper I want in your bibliography, and so it is my first consulted source now. - Prof. Don Loveland (Duke)

Index: I'm really impressed. This is one of the best indexing jobs I've seen in a book in years. There probably hasn't been an index this good since they stopped being done by the proofreaders (back when people still actually proofread stuff...). - Prof. Bob Hobart (Southern Technology College)

Web site: I love your Web page -- I can't believe how many resources I found by glancing over it for a few minutes! - Prof. Jim Schmolze (Tufts)