Useful Publications on AI

I don't have many books on game AI, actually. Most of what I know is self taught through hard experience and experimentation. The web is, frankly, the single best source of information on this topic--no group of dead trees is ever going to be able to match it for sheer volume of information. I have picked up a few tomes that I've found useful, however, and they're listed below. I've found each of these books to be of interest and use. I recommend them all.

I've also come across some other book lists that may be useful; they're listed below.

If I missed anything, please let me know.



o Bryan Stout's Book - Researcher and game developer Bryan Stout is working on a book dedicated to nothing but pure Game AI. He's been seeking input from developers since the 1997 GDC and as of Spring, 2001 it had reached the review stage. Bryan is hoping for publication in 2002.

If you'd like to know more, drop Bryan a note and I'm sure he'd be happy to answer any questions. This book will be fairly ground-breaking, as there's nothing else quite like it out that I'm aware of--let's make it the best it can be!

o Will Dwinnell's Book List - Researcher and author Will Dwinnell has compiler his own list of books on AI theory. It's extraordinarily complete and quite thorough; there are very few references I'm aware of that aren't on it. You can download a Microsoft Word version of it by clicking on this magic spot.

o Game Developer's Magazine - The closest thing to a "Bible" the game industry has, this monthly magazine by Ziff-Davis is aimed squarely at game developers. They occasionally run some interesting articles on Game AI, which is why it's listed here. They have a solid web site where you can read some of the articles and try for a free subscription.

Definitely recommended.

o Steve's Recommendations - As mentioned above, I don't have many books on game AI. I have found these rather useful in one way or another, however.

cover
Get it from Amazon...
TITLE: AI Game Programming Wisdom
Charles River Media - ISBN (1584500778) (planned for March, 2002)
Editor: Steve Rabin

DETAILS: This is the followup to Game Programming Gems and Game Programming Gems 2, and it's a doozy. This one is pure game AI and nothing but.

The format is somewhat similar to the previous two books, though individual articles are a bit longer since we all got more room to breathe. There are 71 chapters by 48 authors in the final version, and it weighs in at a bit over 650 pages. You can find out more about the chapters and such over at the web site.

As before I freely admit to bias here...I'm doing a chapter on Recognizing Strategic Dispositions, sort of a followup to the this excellent thread from several years ago.
cover
Get it from Amazon...
TITLE: Game Programming Gems 2
Charles River Media - ISBN (1584500549) (August, 2001)
Editor: Mark Deloura

DETAILS: Following on the heals of the best-seller Game Programming Gems, this book picks up where the last left off. Bringing more and shorter gems than the first book, GPG2 features 72 gems from some 50 authors.

The new book has been reorganized a bit from the first. The AI section in particular is significantly larger than in the previous book, and section editor Steve Rabin has done a great job of putting a common "voice" to the articles. My own article this time around is a followup on he flocking article I did for GPG1 entitled "Flocking with Teeth: Predators and Prey".

As before I freely admit to bias here, but you won't go wrong buying this book. There's a ton of good stuff here. I hope you enjoy it!
cover
Get it from Amazon...
TITLE: Game Programming Gems
Charles River Media - ISBN (1584500492) (August, 2000)
Editor: Mark Deloura

DETAILS: Written to the style of the classic series beloved by graphics programmers everywhere, Game Programming Gems aims to bring the same kind of practical "this is how you do it" functionality to game programming in general.

Now, it's not all game AI stuff, but there's a big section on the subject-- at least 8 of the 78 Gems are solidly AI. And no I didn't do all of them...my own contribution is on flocking, while AI expert Eric Dybsand did a Gem on state machines. Author Bryan Stout did an A* Gem. Etc., etc.

Yes, I'm biased, but I think this is a good book for game programmers in general, and it's got more of an AI focus than many. If you find it useful, I hope you'll pick it up.
cover
Get it from Amazon...
TITLE: AI for Computer Games and Animation: A Cognitive Modeling Approach
AK Peters, Ltd. - ISBN (1-568-81103-9), August, 1999
Author: John Funge

DETAILS: Author John Funge has done several presentations at various Game Developer's Conferences over the years, but I only recently discovered that he'd also written a book on the topic. John takes a more scholarly approach than many, but covers a wide variety of topics ranging from "Embedding Goal- directed Behavior" to "Hierarchical Finite-state Machines".

The AI concepts are presented in the context of three AI problem exampls-- a maze-solving problem, a dinosaur world in which a big T-Rex tries to eat little Velociraptors, and an undersea world filled with mermen and sharks. They're all pretty interesting scenarios and relate well to a variety of "real world" game development problems.

AI for Games and Animation isn't quite the text for budding AI deveopers, however. John dives into the "math" side of AI very quickly and never really lets up, which can be overwhelming for somebody who's just trying to figure out where to start. Many of the concepts presented are, arguably, of more academic interest than "real game AI" oriented, but that's something that each individual will have to decide for themselvs.

I've found this book to be a useful reference and recommend it if you can handle the math.
cover
Get it from Amazon...
TITLE: AI Agents in Virtual Reality Worlds
Wiley & Sons - ISBN (0-471-12708-6), July, 1996
Author: Mark Watson

DETAILS: This is a great reference on developing interesting AI. Aimed mostly at Virtual Reality games, developer/author Mark Watson has done a marvelous job of bringing together and describing various AI techniques that help make characters in Virtual Reality more lifelike and realistic. Mark covers a wide variety of topics from neural networks to genetic algorithms, then provides a nice set of C++ templates for his AI package. Finally, he incorporates his AI templates into an Asteroids-like game which uses genetic algorithms to "evolve" the playing style of the enemy ships. Good stuff.

Note that Mark also maintains a great web page that focuses on his books, provides source code, and other cool stuff. Definitely recommended.
cover
Get it from Amazon...
TITLE: Developing Games that Learn
Manning Publications - ISBN (0-13-569617-8), 1996
Authors: Leonard Dorfman and Narendra Ghosh

DETAILS: This book is exactly what it says it is--a treatise on building games (and game AIs) that are capable of learning from their mistakes to become better players. The games used to illustrate the various learning techniques experimented with are Tic Tac Toe and Drop Four, fairly simply boardgames which nevertheless (particularly in the case of Drop Four) require both planning and forward thinking. The techniques presented in the book start with what the authors call "dumb" learning--simply not repeating mistakes that led to a loss in a previous game--to an interesting forward-looking approach they call the three-filled marker disruption algorithm.

If I have one complaint about the book it's the source code listings in the book itself. Though the code is provided on disk the authors replicated most of it in the book proper, resulting in an awful lot of pages that I would rather have seen devoted to expanding on the techniques described. Other than that, this is an interesting book and definitely more accessible than some of the more scholarly books currently available.
cover
Get it from Amazon...
TITLE: Evolutionary Computation: Toward a New Philosophy of Machine Intelligence
IEEE Press - ISBN (0-7803-1038-1)

DETAILS: This book is an introduction to the science of evolutionary computation-- the place where genetic algorithms, neural networks, and fuzzy logic combine. It approaches the field of evolutionary computing (and its offshoot, A-Life) from a practical overview perspective, moving from theory to implementation to practice as necessary. A good overview book that explores and explains the topic well.
cover
Get it from Amazon...
TITLE: Genetic Programming: An Introduction
Morgan Kaufmann Publishers - ISBN (3-920993-58-6)
Authors: Banzhaf, Nordin, Keller, Francone

DETAILS: This is a pretty solid book on the subject of genetic programming and genetic algorithms (GP/GA), doing of a good job both of explaining the basics to those new to the subject while providing a useful tool to the expert. There are lots of references, examples, and snippets of psuedo-code, and while there's no CD included with the book there is a web site from which you can download a nicely documented GP/GA software package.

A good place for anybody just learning about GP/GA to start. Note that this book is out of print, so you may have a hard time finding it.
cover
Get it from Amazon...
TITLE: Introduction to Artificial Intelligence and Expert Systems
Prentice Hall - ISBN (0-13-477100-1)
Author: Dan W. Patterson

DETAILS: This book is aimed at beginner, primarily, but it's a good reference for anybody (in my humble opinion). Very throrough discussions on Expert Systems (arguably the simplest and most common form of game AI) and pattern recognition. No real code, but lots of psuedo-code examples.

A good place for an AI newbie to start. Note that this book is out of print, so you may have a hard time finding it.