The AI keeps track of cards discarded, discarded cards drawn, and how many cards are left in the deck, among other things. This allows it to make educated decisions about what to do next based on a utility scoring system. This approach has been very successful against the “test Dummy” DLL we were given to start, often receiving 2-3 times the total end score when playing one-on-one, with a smaller margin but still consistent win rate against larger pools of opponents.
Our competition for the Minesweeper AIs has concluded, and the results are in!
My plan for Minesweeper is to follow the general approach outlined by Bai Li (luckytoilet) in the blog post “How to Write your own Minesweeper AI” which is also linked to in the references section.