I've got a copy of Ian Millington's AI book that I need to finish reading. So far it seems to be very good. It's not as example-driven, but it seems to have good breadth on real-world game AI topics.
The Game Programming Gems series is worth reading through, if, as others have noted, a little dated in places. The general game design and AI sections hold up better than some of the high-performance math and graphics tweaks included from that era. I believe a good portion of them are available for free online, although I may be confusing one of the other Gems series.
I'd recommend them -- most articles are still very relevant. Although they do vary in quality, there are some great reads.
EDIT: Steve Rabin (the editor) has a very high rated new book out called "Game AI Pro: Collected Wisdom of Game AI Professionals" -- http://www.amazon.com/Game-AI-Pro-Collected-Professionals/dp...
The "AI Game Programming Wisdom" books I own the first 2 volumes of. They're the same format as the "Game Programming Gems" (or "Graphics Gems" before that). That is, a series of articles written by many different authors on a smorgasbord of topics. Pros : Wide and sometimes highly-specific coverage and many code samples. Cons : Can be heavily cherry-picked in topic choices. May not be pertinent to anything you need specifically. Refer to the TOC on each respective volume to see if any of the articles pertain to what you need.
Also, this article from Game Developer Magazine offers a cursory introduction to the approach taken by the Norvig book; Agents : https://www.scribd.com/doc/251684769/Adding-Planning-Capabil...
I was so scarred by my godawful, incredibly academic, college AI course that I only remembered the bullshit that we actually covered in that course. Literally nothing in the course was worth wasting my time on, and it looks like we went purposefully around anything that might have been relevant or interesting.
It's baffling to me that two of the worst three computer science courses I took were AI and Algorithms & Data Structures. Both had mindbogglingly abstract, mathematically-oriented profs, who spent every lecture transcribing proofs from the textbook onto the blackboard. How do you teach a course on these subjects and have no assignments that involve writing code? This is at the place where BASIC was created, for chrissakes.
---
The AI Game Programming Wisdom and Game Programming Gems are not revision of the same things. They're just part of the same series; the format of the books is more akin to a collection of articles (all by different authors from the industry), grouped by topic. For example in volume 1, you have:
- General Wisdom (4 articles)
- Useful Techniques and Specialized Systems (8)
- Pathfinding with A* (5)
- Pathfinding and Movement (6)
- Tactical Issues and Intelligent Group Movement (6)
- General Purpose Architectures (6)
- Decision Making Architectures (5)
- FPS, RTS, and RPG AI (6)
- Racing and Sports AI (8)
- Scripting (7)
- Learning (10)
(the groupings are different for each books)
I quite recommend these books - the quality of the articles can be uneven, but you almost always learn something. The main critics against them are that 1) the material can be dated (these books are from the early 2000s) and 2) some of the content can be found online for free. They're still good resources to have on your shelves IMO.
In the same vein, I recommend the GPU Gems and Graphics Gems for realtime graphics programming.
---
"Programming Game AI by Example" by Buckland is a great, hands on, practical approach to game programming. It is not academic or theoretical by any means, but that's one of its strong traits - it's code driven. It was my first AI read while in high school, and I really recommend it. The perfect audience for that would be a motivated teenager or undergrad who knows enough programming to make small games, but hasn't taken full AI courses yet.
---
"Artificial Intelligence: A Modern Approach" by Norvig et al. is kind of the CLRS of AI - a thorough, deep introduction to the field that will give you all the tools you need to solve pretty much any problem you can encounter. It can be a bit dry/academic at times, but that's the price you have to pay. It's great to learn from it as part of a focused course while you're still getting familiar with the field, and it's also great to flip to a random chapter every once in a while to revisit the basics when you're a more mature professional.
---
"Designing Sociable Robots" by Breazal is a cool read - it's a collection of 13 fairly independent chapters by the author which mash together HCI, social sciences, robotics, electronics, programming, and each explain how a subcomponent of a full "sociable" robot (in this case her lab's project, Kismet) works (e.g. the vision system, an expressive vocalization system, etc.).
---
"Society of Mind" by Minsky is a very interesting read. It's more high level and philosophical, but it's definitely worth it. It's almost 30 years old though, and I'm not sure how much his work has held up to modern theories - I don't follow that field too closely.
---
"Behavioral Mathematics for Game AI" by Mark is not that code heavy compared to the others (especially the Game AI Wisdom series), although it does have some, and is rather more oriented towards mathematical modeling. It touches on game theory, utility functions, probability distributions, etc. I think it's the most mathy an AI text can be before becoming pretty much pure math, which is great- it's a great stepping stone for the folks who are programmers first and who want to know more of the math theory behind it all (whether just for understanding or to be able to build their own algorithms).
I am currently reading Society of Mind, and there is defintley still value in the book, there was a recommendation I read somewhere (probably HN) that his newer book The Emotion Machine makes a good follow up to SoM to cover some more recent ideas. That is what I plan to do.
Specifically are the "AI Game Programming Wisdom" books, the new revisions of the same thing? or they complement each other?