Artificial Intelligence/Machine Learning: Program to learn how to play a game.

Title:

Programming Artificial Intelligence for a Computer Game

Objective:

To create a computer program which will be able to 'learn' how to play a game. At 
first it would know basically nothing about the game, and just make random moves, 
but over time it would learn how to counter its opponent, and make intelligent 
decisions.

Justification:

Practically, this is something that has interested me for a while, and after taking the 
Artificial Intelligence class, I want to take it one step further and actually program 
simulated learning. Theoretically, I don't think it has been done before, and it would 
be awesome if I could get it to work.

Description:

I'm not exactly sure which game I'm going to use first, I want to do a somewhat 
complex game to make the project more interesting. I plan on first starting with 
tic-tac-toe, or checkers, or something along those lines, and if that takes me a 
while then I will just stick with that for the rest of the project. If I can finish it 
quickly, then I'll try to move on to something like chess, or maybe some turn-based 
strategy game. It will be faster the second time even if it's a more complex game, 
because I can use much of the same ideas behind the learning.

Limitations:

The only limitations on this project are whether or not I will actually be able to get 
the computer to learn. I may have to do some sort of simulated learning, where the 
computer already kind of knows how to play the game, and just learns how to 
counter strategies, with some knowledge already built in, but I would much rather 
make its knowledge base completely empty at the start.