I'm with Sargas on this. It has taken years for teams of engineers to develop AIs and machines that can play chess to a high human standard. It is possible that you could write a very simple chess controller AI that 'thinks' only 1 move ahead, but I expect that designing an algorithm to (for example) permit the loss of a bishop so that in 3 moves the AI might take the opponent's queen would be incredibly difficult. It's something I would also love to be able to do, but chess is such a complex game (and satisfying for it) that you would probably have to spend years tweaking the factors in your algorithm so that it played reasonably well.
I would start with a draughts/checkers game, just to learn where some general boardgame pitfalls might lie and then maybe go for something like chinese checkers.