How have algorithms advanced chess theory?
Sort:
The way AI learns is by playing lots of games, and if they lose after playing a certain opening, they’re less likely to use that opening again. If they win using a certain opening, they’re more likely to play that opening again. So the choices made by an AI on which opening to play basically tells you which openings are more likely to lead to wins.
Dear chess community,
It is no secret that machine's chess playing ability surpasses that of humankind. The latest development is, that machines teach themselves (reinforcement learning), not utilising knowledge and theory that has been so cleverly devised by us. Apparently, today's strongest chess player, AlphaZero, can teach us something about our theory by playing or not playing certain openings. Humans figured out themselves, that the Prussian opening is a suicide mission and should not be played. Interestingly, AlphaZero seems to have figured out, that Caro-Kann and French might also be be bad. I find this intriguing. Please share some resources and thoughts about theory advancements made or being enabled by machines.
Cheers
P.S.: Check this out: https://en.chessbase.com/post/the-future-is-here-alphazero-learns-chess