In today’s day and age, computers are obviously way better at chess than humans, and even the best players in the world have a hard time holding a candle to engines like Stockfish, etc. However, what I don’t understand is how is it possible to have them in different levels of strength. For example, on Chess.com, there are dozens of bots that you can play against depending on your own level. But for the weakest ones, how does this work? One would think that an engine either knows how to play chess efficiently, or it doesn’t. How can you “dumb down” a computer to the level of an intermediate player or even a beginner? Thanks!
In: 15
The easiest option is to every so often, randomly pick a bad move. But this isn’t realistic, that’s not how bad humans play.
There are certain moves that bad human players don’t see, like bishops on the other side of the board, discovered attacks, pins and skewers. But to a computer, losing a piece is all the same, no matter the type of attack.
So you need to reprogram the chess engine to ignore attacks that are the type of attack a bad player wouldn’t see. This way the human player will actually learn chess strategy.
Maia is a chess engine designed to play poorly. The authors downloaded 10 million games played on lichess.com by human players that aren’t that good. The engine was trained on those games, so it looks for similar situations and does what a human would do.
But, if you want to get better at chess, you really need to play against humans.
Latest Answers