4 ms·
Interesting. I think though the difference with chess is there's no simple known heuristic to evaluate a board position. AlphaGo does it with a neural network,
by hasenj 11y ago
Interesting. I think though the difference with chess is there's no simple known heuristic to evaluate a board position.
AlphaGo does it with a neural network, and as far as I can tell, it seems to be doing a better job at it than top professionals.
All the AIs before AlphaGo basically (as far as I can tell) had some hand-coded heuristic rules.
- V-2 11y ago"the difference with chess is there's no simple known heuristic to evaluate a board position" While it is indeed hand-coded in most chess engines (I'm sure there are some experimental ones, but that's not the path that leads to beating human grandmasters), these heurystics are everything but "simple".