I might be able to help with the AI part, using alpha-beta pruning to estimate future moves to counteract where the player might go. For the difficulty slider it could have a depth slider to tell how many future moves the program should simulate. I've done this before in SmileBASIC