Yeah, I was thinking about (some different dimensions of) this too. Like, humans are better at chess in that chess engines couldn’t abstract gracefully (without extra training) to kung fu chess, or Chess Evolved Online, or other arbitrary rule variants. They’d be especially bad at flick chess.
Also, I think the best Go AIs specifically are reliably fooled by certain wacky tricks that wouldn’t work on humans. Cleanly superhuman performance is hard to find!
Yeah, I was thinking about (some different dimensions of) this too. Like, humans are better at chess in that chess engines couldn’t abstract gracefully (without extra training) to kung fu chess, or Chess Evolved Online, or other arbitrary rule variants. They’d be especially bad at flick chess.
Also, I think the best Go AIs specifically are reliably fooled by certain wacky tricks that wouldn’t work on humans. Cleanly superhuman performance is hard to find!