Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

AlphaGo beats Lee Sedol again in match 2 of 5

gogameguru.com

Post reply on HN
Print view View original

311–320 of 576 posts

Re: AlphaGo beats Lee Sedol again in match 2 of 5

#311

Post by germanier »

philh wrote: post #306
colanderman wrote: post #268

Earlier quoted context omitted.

About a year ago I wrote an AI to play the board game "Hive" (shares some similarities with chess). Because I scored all wins equally, it behaved almost exactly like this. It would simply try to minimize my advantage while always keeping open the possibility for it to win, almost like a cat toying with prey. It never actually would make the winning move – however obvious – until it had no other options! I fixed this…

> It never actually would make the winning move – however obvious – until it had no other options! I'm confused. Why would 'make the winning move' not be the way to maximise probability of winning?

That happens if the winning probability of the other move is 1 as well.

Re: AlphaGo beats Lee Sedol again in match 2 of 5

#312

Post by reacweb »

versteegen wrote: post #200
argonaut wrote: post #183

Earlier quoted context omitted.

I fear this is an overly mystical misinterpretation of neural nets. (Standard feed-forward) neural nets are layers of non-linear feature transformations. They take inputs and at each step transform those inputs into a more compact representation, distilling the inputs into their most important factors (and throwing away unimportant factors). So the most likely explanation is that policy/value nets in AlphaGo have lea…

> So AlphaGo has taken amateur-pro-level board analysis ability, and combined it with tree search I don't think that's an entirely accurate way of putting it, because a player who reaches that level is also doing a little tree searching. Maybe if you found some human who managed to reach "amateur pro" level by playing purely on snap, instinctual decisions without any logic or exploration of variants at all, yes, then…

In go, the tree searching is named "reading". Even beginners need to read very deep trees (more than 7 moves) to anticipate the result of a Semeai (capturing race).

Re: AlphaGo beats Lee Sedol again in match 2 of 5

#313

Post by lukev »

blacktulip wrote: post #282

Does anyone notice the lack of ko[1] in the games? In all 7 public games (5 with Fan and 2 with Lee) there isn't any ko. This is unusual. If we still can't see ko fights in the following 3 games...I would suspect that AlphaGo isn't able to handle ko well enough yet, and Google asked Lee and Fan to not initialize ko fights in the games. [1] https://en.wikipedia.org/wiki/Ko_fight

Isn't it more likely that the opposite is true, and Ko is the kind of thing where it is possible for a computer to play flawlessly?

I am very very very far from a Go expert, but Ko did always seem to be "first person to make a mistake looses". If that's true, I'm sure Sedol deliberately stayed away from it as an area where the computer is likely to be particularly strong.

Re: AlphaGo beats Lee Sedol again in match 2 of 5

#314

Post by edko »

saalweachter wrote: post #286
cubano wrote: post #206

Earlier quoted context omitted.

> Now imagine the winning author of the next Hugo Award turns out to be an AI, how unsettling would that be. I think that would be pretty awesome and amazing, to be honest.

I think it'd be terrifying. Imagine the best book you've ever read. Entrancing, enlightening, cathartic. You reach the end, and it's ... perfect. Oh hey, a sequel. Wow, the sequel is just as good as the first book. It expands upon it without diminishing the original -- you feel better, more complete for having read it. Wait, is that a third book in the series? Wow, it's even better than the first two! A fourth -- wel…

What you describe is a push situation: the books come out and you have to try to keep up to speed with their release. It could however, also be a pull: whenever you feel like reading an amazing book, you just ask the AI to generate one for you, optionally continuing the last story you read.

Re: AlphaGo beats Lee Sedol again in match 2 of 5

#315

Post by awl130 »

philh wrote: post #306
colanderman wrote: post #268

Earlier quoted context omitted.

About a year ago I wrote an AI to play the board game "Hive" (shares some similarities with chess). Because I scored all wins equally, it behaved almost exactly like this. It would simply try to minimize my advantage while always keeping open the possibility for it to win, almost like a cat toying with prey. It never actually would make the winning move – however obvious – until it had no other options! I fixed this…

> It never actually would make the winning move – however obvious – until it had no other options! I'm confused. Why would 'make the winning move' not be the way to maximise probability of winning?

maybe he's defining a "winning move" as something with > 50% chance of winning

Re: AlphaGo beats Lee Sedol again in match 2 of 5

#316

Post by JulianMorrison »

ghaff wrote: post #293
daveguy wrote: post #231

Earlier quoted context omitted.

And, yes, computers can easily smoke all players. FTFY In chess the top engines are rated hundreds of ELO points above Magnus Carlsen (top human). No top ranked human vs computer match has been publicised in over 5 years because humans are thoroughly trounced. There are cyborg matches which are interesting. Human + Computer vs Human + Computer because gameplay techniques are considered different. Humans still depend…

I guess it's because Go isn't that well known in the West but I find it a bit surprising (though not really) that this isn't getting more press. When Kasparov lost it was news but not really surprising. If it wasn't Kasparov that a computer beat, it would have been the next champion. The writing was on the wall for a long time. It was just a question of when exactly. As for Go, I guess I would never had made a long b…

It's probably due to the relatively recent fact that neural networks stopped being "a 1990s fad" and became a thing again.

Re: AlphaGo beats Lee Sedol again in match 2 of 5

#317

Post by angch »

Cookingboy wrote: post #116

Someone somewhere asked why a lot of people in the Go community is taking this in a somewhat hard way, here is my hypothesis: Go, unlike Chess, has deep mytho attached to it. Throughout the history of many Asian countries it's seen as the ultimate abstract strategy game that deeply relies on players' intuition, personality, worldview. The best players are not described as "smart", they are described as "wise". I thin…

Speaking of Hugo Award, all these reminds me too much of Iain Banks's The Player of Games.

An outsider, new to the game, had managed to pick up and challenge top players successfully in a venerated game.

The reactions of community to this is uncannily similar.

Re: AlphaGo beats Lee Sedol again in match 2 of 5

#318

Post by VikingCoder »

colanderman wrote: post #268
astrofinch wrote: post #257

Earlier quoted context omitted.

There's also the fact that some of the unexpected moves were apparently more about solidifying against a loss than increasing the magnitude of a win. Which has its own kind of eerie implication: since AIs (like all computer programs) do what you say, not what you mean, the "intelligent species" can sometimes work really intelligently towards a goal that wasn't quite what you had in mind. (Gets especially interesting…

About a year ago I wrote an AI to play the board game "Hive" (shares some similarities with chess). Because I scored all wins equally, it behaved almost exactly like this. It would simply try to minimize my advantage while always keeping open the possibility for it to win, almost like a cat toying with prey. It never actually would make the winning move – however obvious – until it had no other options! I fixed this…

Thank you for this.

Your post should be required reading in this discussion.

People forget how literal computers are.

Re: AlphaGo beats Lee Sedol again in match 2 of 5

#319

Post by kailuowang »

ganwar wrote: post #210
kailuowang wrote: post #203

Earlier quoted context omitted.

> turns Go into a simple game of mathematics This can be claimed to be true when we understand how deep neural networks mathematically.

Even though we do not have a complete understanding of exactly how the networks work, what is the function that they are minimizing but what we do know is that it is a mathematical function i.e. it has been mathematically modeled. So, I think it is safe to say that it does turn it into a game of mathematics.

Yes the function themselves are pure mathematical functions, but the way to derive them relies on human intuition (not mathematical formula). It's like saying: I know how to do a 1 times 3, I just do 1 + 1 + 1, but I can't tell you why, then I can't say that 1 times 3 is just mathematics.

Re: AlphaGo beats Lee Sedol again in match 2 of 5

#320

Post by typon »

aaimnr wrote: post #295
typon wrote: post #280

I've had this thought watching this play out over the past few months. You have this deeply mystical, zen-like game of ancient China which represents the philosophy of the East and it's pitted against this pure capitalist, cold and calculating (literally) machine. You can hold out for a few thousand years, but eventually the uncontrollable and amoral technological imperative will catch on and crush you. It's kind of…

On the other hand one could say that it's a pretty shallow understanding of Zen, given that Buddhism assumes that everything in the mind - being a part of nature - is fully conditioned (deterministic), as opposed to Christian view of a human as a holy, God-like creature, distinctly different than the rest of nature.

I agree. I think the same concept applies in Daoism. Sort of giving yourself up to the flow of the universe. That's why I only said "zen-like".
Post reply on HN