We just finished covering Neural Networks in my Artificial Intelligent class and I've come to the quick conclusion that they are totally radical. Unfortunately, neural networks are in fact too radical and are rarely used for games - the AI gets too good at the game much too fast.

That kinda bummed me out, so I started thinking about a way that I could use neural networks in game design...

Neural networks happen to be excellent at image recognition; the optimization process lends itself to the task quite well. So my plan would be to use image recognition on a camera that watched the players face so I could track emotion. You could then extrapolate how the player was feeling about the game and adjust the game play accordingly.

I would use the system to create the most frustrating game ever. If you smile then tweak the game to piss them off. You grimace at the first jumping puzzle - oh uh the next stage is ALL jumping puzzles! Take that! Don't let the game know that you can't stand swimming areas or else the city might flood.

A nice touch would be to have a accelerometer built into the controller to detect the point at which it went flying across the room (the ultimate form of gamer frustration) so that the Game Over screen could pop up and mock the player accordingly.