The human was given a 2-stone handicap.
Shin played with an advantageous, two-stone handicap—seemingly standard for matches between humans and top AIs
Apparently that’s the norm, and humans have still lost up until now.
The headline led me to assume that they were scratch games until I read the article. Still, an amazing performance when you consider that it’s been a decade since Lee Sedol lost to AlphaGo.
Im pretty sure computers/AI started beating the top humans at go at least like 10 years ago. It definitely was well before covid and the current AI boom. Although Im sure the computers have only gotten better since then
Source? My recollection at the time was quite the opposite; that pros consistently found AI were weak in large sections of the end game. The lookup tables required to do what stockfish does for chess are absurdly large, iirc.
Alpha go won 4 to 1 in 2016. Might have passed you by or it seems closer in time but ten years was spot on.
Ahhh; I was not reading carefully enough. You are right.
I was, for whatever silly reason, reading “current AI boom” as “current AI methods”; AlphaGo was a neural network model, very close to what we do for LLMs, genAI, etc. My impression was that before NN’s, we had a good lead over computers.
I didn’t know Go has grandmasters
9 Dan
Next they should try a young girl who is blind and has only played Go her whole life. I bet she’d win without a stone handicap.
I’ll consider it when I get permission to have a racoon as pet which has a tail ending in a syringe tip.
Removed by mod
Humans invented and made the AI, so idk why this article is phrased as if humans are inferior :D
The thing the AI can do is consider 14 million futures and choose the one where it wins. Humans usually can’t.
But humans can jump to conclusions that AI would never “think” of.
actually recentish chess bots showed playing in a uniquely bot sort of way by sacking points for long term position control well past current human norms was a viable strat
google ‘deep positional sacrifice used as a tool for total restriction’ for videos on it
so if anything it’s reverse
chess isn’t solved but it is deterministic not some purely subjective creative art
why would a human be better at it?
Theoretically the humans would win one every so often because sometimes humans “zig” when AI can only “zag”? It’s like scientific calculators doing math way more quickly than any human on earth, but they are not necessarily better at it. I don’t know, it’s a brave new world and you may be right.
yea nah
it being a deterministic game means there is a right answer or correct play at each point
now we obviously realized our foundational principles are wrong at times, otherwise this new computer method would not have been shocking but that was a human failing and not something you could exploit against a bot
the kind of leap you are envisioning is something that would only really work in a chaotic system
the idea being there that human can leap to a right conclusion not completely supported by facts
you could arguably just call that gambling
I wouldn’t call it a “leap” but a few years ago some researchers did find a weird byway in the then-current go-playing "AI"s (neural-network setups, as someone else pointed out). They had to cheat a bit by examining the NN’s “thinking” more directly, and found a cyclic strategy that allowed a human to beat the machine.
https://www.far.ai/blog/even-superhuman-go-ais-have-surprising-failure-modes
It’s interesting for more “AI”-type stuff generally. The point is not so much “us with our special brains will always find ways to defeat AI” but more “there are odd blind spots that you would not predict by just looking at the output/games, and these can be exploited with appropriate technology.”
Edit: remove redundantly redundant redundancy
that is cool and not uncommon in algos in general
bad cost function, poor fit / overfit, local inflection etc
Yes? Why are you saying this to me?
By that logic all children are inferior to their parents, and therefore the greatest of us is… Oog the caveman.
return to monkeigh
Huh, which logic? I’m genuinely confused. I’m just saying that it’s weird that the article is phrasing it like humans are inferior to the Go engine when it’s literally humans who created the Go engine. Just because humans lose to a machine doesn’t make humans inferior, which the article insinuates.
The implication that something that creates something else is superior to the thing it creates is a fallacy. There is no inherent relationship between them whatsoever - it is only measurable via direct measurement of performance. The claim that “humans created the Go engine makes them superior” is exactly the same amount of logically incorrect as the “it’s a machine so it’s superior” claim.
If your claim that creating something means superiority held true, Lebron James’ dad must be better than him at basketball. And I know that’s a ridiculous claim. But it’s logically equivalent to your own.
Do you really think your parents created the entirety of what constitutes “you”?
It’s not logically equivalent at all, a child learns on its own without parents, or through all kinds of other influences. In addition, if your parents actually knew how to make a human, they would be able to exactly replicate a previous human they created. But this is absolutely impossible, even with genetic clones, as no human actually has any idea how to get another human to be exactly the same as another.
In contrast, someone (or a collection of humans) understands completely how the best Go engine works. They can perfectly replicate the current Go engine by writing the same code again. The Go engine is a product solely by the human mind and, if the code were lost, can be remade by applying the same principles.
“Creating a human” is nothing any human has ever done. Humans have been created by million of years of evolution. In this context, if you want logical equivalency, “nature/universal laws” (or if you’re so inclined, “god”) would be the equivalent representation of what humans are to a Go engine.
We actually don’t understand the inner workings of most neural networks, this is a well-known problem. We can evolve them to solve a problem but we won’t necessarily know how they are solving it.
“Gun kills things faster than a human can, humans now suck”
Yeah exactly, it’s so weird that the article phrases it like that. I don’t understand why I got so many downvotes :D
Most partial differential equations have no analytic solution, but we can still solve them numerically. We design the algorithms, we prove they provide the correct solutions in the right conditions, but given the amount of computation and steps required to arrive at a solution, for the hardest problems (ie navier-stokes), it is practically impossible for a human to do by hand.
It doesn’t mean the computer is better. It’s just a tool used to solve a problem. In this case that problem is go, and that computer is really good at solving it.
Yeah exactly. My original statement was that it was weird that the article is phrased as if humans were inferior.
