As part of my undergrad work, I used similar principles to the article (steady state genetic algorithms) to create a bot capable of playing Reversi where the fitness function was loosely defined as a set of "weights" on each square across the 8x8 board. These were used as part of the evaluation function in the classic minimax algorithm.
We trained over the course of 5-6 days, and the end generation was capable of beating an intermediate player but got completely destroyed by experts. It was a fun project!
replies(1):