If you are going to the time and effort to write a multi-part article on these techniques, then I hope you will discuss the implementation issues of using these techniques in computer game AI, instead of just presenting definitions and descriptions of the various techniques.

Most articles on ANNs, FL and GAs found on gaming sites, simply describe the techniques and gloss over (or avoid) discussing the more critical issues of "How to use an artificial neural net in a game?" or "What can an artificial neural net do for you in a game?" or "How can a genetic algorithm be usefull in computer game AI?" or "Why use fuzzy logic instead of a finite state machine?".

For instance, one would not want to use an ANN, FL or a GA for solving a pathfinding problem in a computer game, when the AStar algorithm has proven itself (via widespread usage, implemenation and practice) to be a much better way to find a path (in almost all types of worlds). Describing "why" the AStar is better than ANN, FL and GAs (for this type of computer AI problem), is more useful than just presenting what an AStar is.

Anyway, good luck on your article.

Geta

I totally, totally agree.
I have read articles in Games Developer on "AI : The state of the industry" and they may be okay to get a vague overview of what AI is, how it can be used and what it can do, but they do little to describe how, when and why to implement specific systems.
Like any major aspect of games programming, once you've made you're decision to implement a specific form of AI to solve a problem, it becomes increasingly difficult to scrap it and restart once you come to the various lmitations of the different methods.
It would be great to think that we have the time and funding to write various testbeds with which to try out different styles of AI to produce the most realistic, quickest, producer pleasing solutions but often there are more limitations than freedoms.
This is where a good reference on guidelines would come in extremely handy explaining where different forms of AI would be the most useful and why some might work better than others in different situations.

This is a great idea for an article. I just hope that it is not a thesis on the theory of Neural Networks. I would love to see implementation suggestions of Neural Networks in Games.