[mjn]

As an AI academic who used to have a bunch of game-AI industry contacts at big companies, my anecdotal impression is that a lot of people are indeed taking this advice. Either that or I'm unlucky, because my contacts are disappearing about as fast as I can make them...

A few are still doing games now, in the indie scene. The most interesting of those, imo, is Richard Evans (formerly of Black & White and The Sims 3), who's building an interactive-fiction platform on top of a novel logic-based social-simulation engine (http://www.versu.com). But a lot are doing entirely different things: some moved to "regular" programming jobs, ranging from simulation programming for big engineering companies or defense contractors (somewhat related) to almost anything at all that provides a regular job. A number have gone back to grad school. A few have become freelance consultants.

Imo it really stunts the growth of game AI as well, because there are remarkably few experienced senior AI devs who actually stay in the field, long enough to develop a sense of how to architect systems, how to interface with designers, and actually improve things over time.