# Good AI Theory Books

## Recommended Posts

I am looking for a book that is more in the theory of AI and the understanding then the underlying programming of it. Since my programming skills are still minimal at best and basic games are the best I can program. However I would like to major in AI and want a head start. So any good books or authors you can recommend would be great. Thank you

##### Share on other sites
"Artificial Intelligence: A Modern Approach" is the most common college textbook on the theory. There is another I like that is simply called "Artificial Intelligence", but it is at work so I can't give you the correct authors right now...

These books will assume that you are already familiar with basic algorithmics, data structures, search algorithms, graph theory and the like, and that you have college-level math skills.

##### Share on other sites
I was going to recommend "Artificial Intelligence: A Modern Approach" as well. That's the textbook we used when I tutored the subject, and it's a good book. However as Steadtler wrote it might assume you know the basics of algorithms and data structures.

##### Share on other sites
There's another popular one called "Computational Intelligence: A Logical Approach".

##### Share on other sites
I have a copy of "Introduction to Artificial Intelligence" by Philip Jackson sitting on my TV. I'm not sure how much programming theory it has, but there is definitely theory, even most of a chapter devoted to axons and neurons and the like. Haven't had time to read it all since it's kinda boring to be honest.
Not a bad read for only about $20 at B&N. #### Share this post ##### Link to post ##### Share on other sites Russel & Norvig, "AI - A Modern Approach" is the de facto standard. Before I got "officially" tutored using this book, I was teaching myself with George Luger's book on AI, which I liked as well. R&N was deeper on theory but - for my perception - occasionally does a bad job at explaining certain things. So I noticed that without my memories of Luger's slightly more comprehensive explanations, I would have had a more difficult time. I'd still go with R&N. #### Share this post ##### Link to post ##### Share on other sites Given the parameters you mentioned (i.e. theory with little or no code) then R&N it is. BTW, it's not small. Put a chiropractor on retainer. #### Share this post ##### Link to post ##### Share on other sites Quote:  Original post by SleepRussel & Norvig, "AI - A Modern Approach" is the de facto standard. I have heard that, despite the name, this book is now a little dated. Can anybody comment on this claim? #### Share this post ##### Link to post ##### Share on other sites Quote: Original post by Kylotan Quote:  Original post by SleepRussel & Norvig, "AI - A Modern Approach" is the de facto standard. I have heard that, despite the name, this book is now a little dated. Can anybody comment on this claim? True, it's been 6 years since the 2nd edition. I am not expert enough as that I could recommend something else, though. In terms of theory, though, not too much should have changed. Search still works the same way, knowledge representation still works on FOL, bayesian networks are still widely used, etc... Of course, Machine learning and NLP of course are fields on their own and make fast-paced progress. If somebody would search up-to-date information on this, I would not recommend a general AI book anyway. Actually, if someone has a state-of-the-art book at hand, I would want to know about it, too. #### Share this post ##### Link to post ##### Share on other sites Quote:  Original post by Steadtler"There is another I like that is simply called "Artificial Intelligence", but it is at work so I can't give you the correct authors right now... I was going to recommend that one, it's written by Luger and published by Addison Wesley. If that's not the one you were recommending, well, I certainly would. #### Share this post ##### Link to post ##### Share on other sites The 2nd Edition of R&N's AI:AMA has significant revisions over the first edition and while it is >5 years since publication, the whole point of this tome was that it would serve as a basis for undergraduate college level courses in AI. It provides you with sufficient material in most of the major areas of AI study to give you a basic understanding of the issues involved, from problem specification through to methods to solve those problems. Obviously research has extended the works in most, if not all of these areas. That's not to say you should not learn what has gone before. AI:AMA is not, by any stretch of the imagination, all you'll ever need to read to understand AI. It's an introduction to the core areas of AI and to date, still the best 'introductory text' I've read. If you have a specific area that you want to pursue there are more specialised and focused books that you should read. From my perspective, buying AI:AMA will never be a waste of money. You may find a day when you stop picking it up (because your knowledge has surpassed that conveyed in the book), but if that's the case, it served its purpose well! Cheers, Timkin #### Share this post ##### Link to post ##### Share on other sites Thank you =) I am currently in Discrete Math 2 (We have gone over most algorithms and we are currently working our way through trees). I have done search algorithms, some data structures and will be doing the programming data structures class next quarter. So hopefully that is enough to understand these books. #### Share this post ##### Link to post ##### Share on other sites Quote: Original post by TheGilb Quote:  Original post by Steadtler"There is another I like that is simply called "Artificial Intelligence", but it is at work so I can't give you the correct authors right now... I was going to recommend that one, it's written by Luger and published by Addison Wesley. If that's not the one you were recommending, well, I certainly would. Nah, its actually by Rich and Knight, published by McGrawHill. Very good for symbolic AI. #### Share this post ##### Link to post ##### Share on other sites There are a couple of books not mentioned yet that you might find useful. Check out Douglas Hofstadter's Fluid Concepts and Creative Analogies. (About$30.) It's a readable, interesting book that describes his research group's efforts to build creativity into machines, in enough detail to be useful but without getting into the exact code. The book also has discussions and critiques of several other approaches to AI. Hofstadter's Pulitzer-winning Godel, Escher, Bach is partly about AI and a lot of other interesting things, but it's not AI-focused and parts of it are dull math theory.

You might also be interested in Chapman's Vision, Instruction and Action, which describes the author's attempt to build an AI demonstrating survival in a Gauntlet-like game world. Too technical to be fun reading, but the parts about the theory behind it are worth reading. Stepping back from AI a bit, I suggest Stephen Pinker's How the Mind Works for some insight on thinking in general. Pinker's work is relevant to AI, and his writing style is admirably clear and entertaining.

I also suggest checking out the e-mail list "Robitron," where Hugh Loebner of the Loebner Prize Contest and various contestants talk about AI. The focus there is on the "chatterbot" approach, which I totally disagree with, but there are still some interesting discussions.