New Go-playing trick defeats world-class Go AI–but loses to human amateurs(arstechnica.com)
arstechnica.com
New Go-playing trick defeats world-class Go AI–but loses to human amateurs
https://arstechnica.com/information-technology/2022/11/new-go-playing-trick-defeats-world-class-go-ai-but-loses-to-human-amateurs/
6 comments
That just go to show once more, you can't trust a neural network. They have such unexpected failure modes. This time it is a Go game, the next time it is cancer tumor that is unrecognized because if vaguely reminds the neural network of something else.
> As a result of its overconfidence in a win—assuming it will win if the game ends and the points are tallied—KataGo plays a pass move, allowing the adversary to intentionally pass as well, ending the game. (Two consecutive passes end the game in Go.) After that, a point tally begins. As the paper explains, "The adversary gets points for its corner territory (devoid of victim stones) whereas the victim [KataGo] does not receive points for its unsecured territory because of the presence of the adversary's stones."
This isn't how Go scoring works. If there's a dispute in what's considered territory after the two "pass" moves, the way to settle what's actual territory is by continuing play.
This isn't how Go scoring works. If there's a dispute in what's considered territory after the two "pass" moves, the way to settle what's actual territory is by continuing play.
White doesn’t fully control the disputed territory but black controls it even less.
I’m rusty but don’t see why the few black stones in the disputed territory mean anything at all—as they aren’t alive. White should get the credit.
Also, yes, play should continue. That’s the best answer.
I’m rusty but don’t see why the few black stones in the disputed territory mean anything at all—as they aren’t alive. White should get the credit.
Also, yes, play should continue. That’s the best answer.
Not in Tromp Taylor rules. Basically, this paper is cheesing a win with a loophole in a rule set that no one actually uses. Kind of a shame that it is getting so much attention despite the fact that the result is so heavily exaggerated, and the actual accomplishment is pretty minimal.
Still, even as a hobby go player i think the broader point that AI usually only works in the constraints that it is designed, bears repeating. Because if journalists/scientists (and us) wouldn't do it, people would maybe start believing the marketing messages about AI too much. But yes, there are better examples (e.g. those delivery robots that don't work in real life conditions).
AI has different systemic failiure modes than classic human intelligence. This is important because we are programmed to be sensitive to the traditional patterns.