3 ms·
The AI can always continue with the perfect play from that imperfect state. That is what happens in chess as you can’t fool a good chess AI by making intentiona
by doikor 3y ago
The AI can always continue with the perfect play from that imperfect state. That is what happens in chess as you can’t fool a good chess AI by making intentionally imperfect moves (chess is not solved so we don’t know if a move is perfect or not but you get the idea)
Basically in games with no hidden information (chess, go, othello, etc) how you got into a game state does not matter. If your AI can play perfectly then it can make the perfect move from any state.
- 542458 3y ago> you can’t fool a good chess AI by making intentionally imperfect moves You actually used to be able to. That’s part of how Kasparov beat Deep Blue in ‘96 IIRC - by forcing the game into unusual situations where the computer would have fewer prior games to pattern match off. But by the rematch heuristics for “off book” situations had improved, and the same strategies failed (or even backfired).