http://www.ieee-cig.org/cig-2009/Proceedings/proceedings/index.html
The interesting thing is that, while game designers are surely more concerned with creating an impression of intelligence than the real thing, the techniques discussed in these proceedings – Monte Carlo optimization, temporal differential learning, evolutionary programming, neural networks – are all developments of "true" research in AI and optimization. It seems, then, that at some point to "fake it" was not enough anymore, and games now have to try and "make it" to keep their pace of advancement. I would even dare to say Game AI contributes to the development of "real" AI, by providing an wide application field where the practical limits of AI techniques are put to the test.
The gaming industry has for some time now driven upgrades in hardware, with increasingly complex games requiring ever-growing memory and processing resources. Perhaps it will also one day drive new developments in software – and particularly AI – programming techniques?
