Very interesting, but doesn't this increase the options for cheaters, if the moves are more human like?
What about playing with a high level maia against a middle level human. Can the cheating detection rekognize this?
@Shogun77 That's what I was thinking too, but I'm sure they'll have some type of way to figure it out.
Maybe all the games on lichess are already just ai and we do not yet know it. Just joking hopefully.
This is absolutely brilliant!
They play so much better than their supposed ELO. Also, to play like a human, they can't make twenty movements in 0.0001 seconds
........ Be that as it may, please stop calling all these bots AI as they are millions of light-years away to pass the Touring test. They are bots, not human-like neural network chess engines ...... They make some mistakes, indeed, but even all these mistakes have been already coded.
That is super cool!
I don't understand the point in making it a chess engine. Isn't it just a data collector? And for the purpose of data analysis, it doesn't need to play chess actually. From the article it seems like that's the main reason. So the fact it plays chess is just a gimmick, that doesn't even help from the perspective of data collection.
I think this is the most exciting thing to happen to computer chess since they overtook humans in strength. I find it interesting that it took so much longer to develop a realistic, weaker opponent, than it did it develop the strongest opponent. I look forward to playing these! I hope they stay online.
Everything’s so great, i really am impressed in how human the moves are. Although i do suggest move times be studied as well because
@maia9 flagged me although i had 25 seconds to 1 :(
Other than that, great stuff!
Perhaps one day the computer analysis of our games will not just say "mistake", "inaccuracy" or "blunder", but also "typical of below-1500", "typical of below-1700" etc. If this could be made accurate, it might enable occasional players to estimate their rating from just one game. (The estimate might be wrong though.)
