Disclaimer: the "AI" was written by my friend Jan, not me, and I haven't dug into that part of the code. I do know there is no deep search of all possible moves several steps in the future. Instead, we were given some basic rules by a professional player (a Belgian Champion in Othello), which Jan implemented, and then optimized. Those rules just look at the current board state, and no more. The different difficulty levels just apply more of these rules, making the AI player smarter.
Viewing post in Grand Reversi jam comments
Indeed. At this moment he doesn't look forward. Only some basic tips you should give to a player. So not using the calculation power of the pc.
I was thinking about the tree that you mentioned. That would actually be easier (less possibilities than chess and max 60 moves). But I was afraid that the game would be barely unplayable.
I have the plan to extend the AI with just that