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

Google reveals secret test of AI bot to beat top Go players

nature.com

Post reply on HN
Print view View original

161–170 of 214 posts

Re: Google reveals secret test of AI bot to beat top Go players

#161

Post by whiteandnerdy »

furyofantares wrote: post #70
dfan wrote: post #7

One thing that isn't made clear in this writeup is that Master plays in a very nonhuman style, as opposed to the version of AlphaGo that beat Lee Sedol, which mostly played like a strong human except for a few surprising moves. My first guess when I saw Master's games was that it was a program like AlphaGo that had its policy network trained from scratch rather than being bootstrapped by being given the goal of imita…

Since AlphaGo's original training was to predict human moves, it should know how surprising a move is in addition to knowing how strong it is. My thought at the time was it could improve its game against humans by giving surprise value some weight -- when it can pick a slightly weaker but highly surprising move, pick the surprise.

It definitely does know how surprising a move is - it generates 'surprisingness' numbers for both itself and its opponent based on how likely it is a human pro would have come up with the move.

I like your idea of maximising the 'alienness' of play though probably not to the exclusion of playing the strongest moves.

Re: Google reveals secret test of AI bot to beat top Go players

#162

Post by tempestn »

furyofantares wrote: post #70
dfan wrote: post #7

One thing that isn't made clear in this writeup is that Master plays in a very nonhuman style, as opposed to the version of AlphaGo that beat Lee Sedol, which mostly played like a strong human except for a few surprising moves. My first guess when I saw Master's games was that it was a program like AlphaGo that had its policy network trained from scratch rather than being bootstrapped by being given the goal of imita…

Since AlphaGo's original training was to predict human moves, it should know how surprising a move is in addition to knowing how strong it is. My thought at the time was it could improve its game against humans by giving surprise value some weight -- when it can pick a slightly weaker but highly surprising move, pick the surprise.

You're probably right, if they were optimizing for short-term win rate vs humans. The real goal is to advance the state of the art in AI though, in which case you wouldn't want to use 'cheats' like that to make the problem easier.

Re: Google reveals secret test of AI bot to beat top Go players

#163

Post by zachrose »

malteof wrote: post #149
inimino wrote: post #141

Earlier quoted context omitted.

Games like go and chess are necessarily solveable by definition, right? If you mean solveable in our lifetimes, by current technology, that's something else entirely, and much less likely. Beating human players in go is like finding a good heuristic for an NP-complete problem. Solving 19*19 go is like proving P!=NP, i.e. we don't even have tools that can approach the problem.

Not sure if this is true for Go. I'm not sure, but it seems to me that at some point a problem becomes "too complex" to solve. I.e. Go has about 2.08168199382×10^170 legal positions [1] and the number of atoms in the observable universe is only up to ~4×10^81. I don't know if there are some theorems, thoughts, philosophies about whether this means it can't be solved, but at least it must be extremely difficult. [1] h…

While mind-boggling, I'm not sure it's generally meaningful to compare counts of actual things (even atoms) with counts of possible arrangements. The latter are in a world of their own[1].

[1] https://en.m.wikipedia.org/wiki/Orders_of_magnitude_(numbers...

Re: Google reveals secret test of AI bot to beat top Go players

#164

Post by corkscrew »

on_and_off wrote: post #158
halflings wrote: post #89

Earlier quoted context omitted.

Directly thought of StarCraft while reading this comment: never revealing your strategy and constantly attacking while being defensing and investing in economy to keep your opponent on the defensive. Like another commenter said, StarCraft 2 is DeepMind's next bet.

I am not too familiar with StarCraft but from what I can gather, it relies deeply on micromanagement. It seems to me that any half competent AI strategy wise would destroy any human player just with raw clicks per seconds during fights.

As far as I'm aware they force the AI to only interact at a certain frequency, limiting this kind of advantage

Re: Google reveals secret test of AI bot to beat top Go players

#165

Post by codehotter »

tux3 wrote: post #4
MR4D wrote: post #2

This is impressive. According to the article, the bot has not yet lost. I think at the end of 2017, we may have to say that computers have conquered Go. Interesting that the timing is 20 years after mastering chess.

It's important to note that the games were played on a fast-paced "blitz" time setting where AIs are particularly strong compared to humans. I can already barely follow what is going on on the board, so I'm sure watching slower paced games against pros would be interesting.

The games were played with 5 seconds per move for the computer and 30 seconds per move for the human. (Where unused time is lost, not saved).

I am not sure if slower time settings favoring the human is still true -- AlphaGo works a little bit differently from traditional computer go programs.

Earlier versions of AlphaGo against Fan Hui, the slower official matches were 5-0 in favor of AlphaGo, whereas in the unofficial blitz games, Fan Hui managed to win some games.

Re: Google reveals secret test of AI bot to beat top Go players

#166

Post by mschwaig »

on_and_off wrote: post #158
halflings wrote: post #89

Earlier quoted context omitted.

Directly thought of StarCraft while reading this comment: never revealing your strategy and constantly attacking while being defensing and investing in economy to keep your opponent on the defensive. Like another commenter said, StarCraft 2 is DeepMind's next bet.

I am not too familiar with StarCraft but from what I can gather, it relies deeply on micromanagement. It seems to me that any half competent AI strategy wise would destroy any human player just with raw clicks per seconds during fights.

That's true, with StarCraft you have to basically limit the actions per minute the AI can perform to human levels, otherwise it can effortlessly out-control a human player. The game was only balanced to be fair for players with human capabilities, if you take that assumption away the unit and race balance goes totally off the rails.

There is the question of what your AI takes as its input though. If you feed it the rendered frames of the games there are a lot of open challenges in comparison to feeding it a friendlier representation of the game state.

There is also the question of teaching the AI a strategy vs only giving it the rules of the game and letting it develop strategies by itself.

In comparison to Chess or Go, StarCraft has a lot more game elements to understand and conceptualize. StarCraft is played with incomplete information, because of the fog of war - so for an AI to play it properly it has to have persistent knowledge of what it has seen of the game state and be able to draw conclusions from that about the probable overall state of the game.

Re: Google reveals secret test of AI bot to beat top Go players

#167

Post by codehotter »

rdlecler1 wrote: post #84

I'd like to see if a small team of the worlds best go players could beat AG.

How much stronger is a team of top players than its strongest member?

I wonder what the best way of coordination would be. Perhaps they can identify several promising lines and each player chooses one variation to calculate more deeply.

From my own experience playing (as a weak amateur), I feel I'm rarely able to think so systematically -- often ideas I discover while contemplating one line are tried in entirely different variations, so it would perhaps take some getting used to before a team can function optimally.

Re: Google reveals secret test of AI bot to beat top Go players

#168

Post by Retric »

tasuki wrote: post #157
Retric wrote: post #134

Earlier quoted context omitted.

Perfect play is likely inhumanly aggressive on blacks part part and white making zero moves. Compared to that this is very human style of gameplay simply based on a different strategy culture as it where.

I don't see why perfect black play should be any more aggressive than perfect white play. Care to elaborate?

Black moves first, on a 3x3 and 5x5 both end up 100% black with any white piece being captured. Many other board shapes don't but Go is played on a 19 x 19 board. We don't know about 9x9 or even 7x7 so the pattern is hardly set in stone. Still it seems likely.

Now, with a perfect white play there may be moves an imperfect black player makes which causes white to attack. But, perfect play on both sides probably means any white stone gets captured so white plays zero stones.

Re: Google reveals secret test of AI bot to beat top Go players

#169

Post by iopq »

aaron695 wrote: post #155
conanbatt wrote: post #127

Earlier quoted context omitted.

As a semi-pro player I can assure you thats not how you play Go.

You never study your up coming opponents previous games?

For example, US Congress has a lot of matches in a day so you don't even know who you're going to face. Only top players study the games of their opponents because they have time to prepare in finals of big tournaments.

Re: Google reveals secret test of AI bot to beat top Go players

#170

Post by nightcracker »

kenshaw wrote: post #136
kirse wrote: post #129

Earlier quoted context omitted.

they'll happily enter lines where their pawn structures are shattered... Makes you wonder if computers will essentially ruin all the novelty of innovating and discovering new tricks/strategies in age-old games. No more will genius' have their moments of brilliance and improvisation that lead to a stunning victory or upset. No more will commentators debate and regal over phenomenal victories in ages past. With a compu…

Probably. I think it basically proves that any rule structured game without randomness is solvable.

There is a trivial disproof of your claim.

Two players write a turing machine with at most n states with two symbols. The player that produced a terminating turing machine that produces the most 1 symbols on the tape before terminating wins.

The optimal strategy for this game is producing a busy beaver, a feat shown not to be computable.

Post reply on HN