Google has beaten Go, what could it do to bridge?

[nullve]

It is true that bridge includes some aspects that chess and go do not. Most notably, incomplete information and partnership communication. Against this, the game space is just much smaller: fewer decisions overall, fewer legal alternatives per decision, and fewer reasonable alternatives per decision.

But which game are we talking about? If it's the game that begins only after systems have been designed and agreed upon, then the game tree is finite, just like in chess. If it's the game that starts with system design, then the game tree is infinite, because e.g. for each p in the real interval [0,1] a pair could choose a system where hands of type T were opened 1C and 1D with probability p and 1-p, respectively.

[Fluffy]

Even if an AI plays perfectly and has a perfectly designed system they would still be underdogs to win BB. They would probably win the RR by a large margin, but then the KO would be a big problem.

[mike777]

Even if an AI plays perfectly and has a perfectly designed system they would still be underdogs to win BB. They would probably win the RR by a large margin, but then the KO would be a big problem.

 

If, I mean If, AI plays perfectly they will win the Bermuda bowl today, AI will win even if they play less than perfect, if not today then before 2050.

 

btw I understand AI has won the first two games out of 5

[yunling]

There was a match in 2005 between Jack and some top Dutch pairs

http://www.jackbridge.com/eperhum.htm

 

But I can't find any report since.

Probably it is because computers find it very hard to understand humans' alerts. :unsure:

[lycier]

The fouth round, Li se-dol wins, congrats !

Lee se-dol standing on the intersection of the present and future ,just like a solemn and stirring warriors !

 

http://i63.tinypic.com/25z095f.jpg

 

 

http://i66.tinypic.com/nb4to3.jpg

 

 

http://i68.tinypic.com/2iha6j9.jpg

 

 

http://i67.tinypic.com/316rlz8.jpg

 

[jogs]

The fouth round, Li se-dol wins, congrats !

Lee se-dol standing on the intersection of the present and future ,just like a solemn and stirring warriors !

 

http://i63.tinypic.com/25z095f.jpg

 

 

http://i67.tinypic.com/316rlz8.jpg

Is that a chess clock? Lee was down to 1 minute left. A human opponent would have tried to force Lee to lose on time.

[mgoetze]

Is that a chess clock? Lee was down to 1 minute left. A human opponent would have tried to force Lee to lose on time.

Leaving aside some minor details, the players essentially had one minute per move after the main time (2 hours) expired. People don't lose on time when they're ahead.

[lycier]

I am sure the people behind AlphaGo would love to take him on. Ke Jie himself might be a little more reluctant having more to lose from such an encounter. You can be sure he is studying the computer's play very closely looking for weaknesses.

 

http://i68.tinypic.com/oh09d.jpg

 

 

http://i63.tinypic.com/28gv9jd.jpg

 

 

 

 

 

I am sure that Ke Jie is looking forward to challenge AlphaGo.

[mike777]

First three=AI

Forth=Human: Human says they won by playing unorthodox.

--------

 

Many Go players complain Human is old and weak player and should retire.

[Zelandakh]

Human says they won by playing unorthodox.

WHich is precisely how it worked for a while in chess. Computers had difficulty with very quiet positions with few tactical possibilities. Once strong GMs got on board in closing this hole, computers were able to hold their own in those positions too. Having exposed this weakness now though, Lee has perhaps paved the way for other human Go players to best AlphaGo in the short term. It takes time for a program to be re-worked and until that happens there is a window for Ke and the others. It should be clear from the chess history that that will not last though. It should also be clear that players that use the tactic discovered by Lee in future matches to beat AlphaGo are not stronger - the key is finding the weakness and then having enough ability to exploit it.

[lycier]

Today in final round, AlphaGo have the victory over Lee sedol, the total score : 4:1

Congratulations to Google AlphaGo team ! Hats off to DeepMind founder Demis Hassabis !

 

1- Li kaifu thinks the thinking ability of AlphaGo on playing Go is stronger than human.

2- Almost of the professional Go champions admit AlphaGo's skill is stronger than human.

3- I think DeepMind founder Demis Hassabis, UK scientist, should obtain the Nobel Prize.

 

In weibo, more than one hundred million people are paying attention to the event. Some players are hoping that AlpfaGo may come out unexpectedly bug again, great pity that AlpfaGo never give any chance to human.

All the professional Go players of China are looking forword to play against AlphaGo. Our chinese desire is Welcome AlphaGo and Demis Hassabis to China !

 

http://i64.tinypic.com/2mmvqs.jpg

 

http://i68.tinypic.com/s1124x.jpg

 

http://i64.tinypic.com/2iqzdif.jpg

 

http://i65.tinypic.com/2zji4n5.jpghttp://i63.tinypic.com/x1acgn.jpg

 

 

http://i64.tinypic.com/2ujm5iu.jpg

 

http://i68.tinypic.com/2h8702g.jpg VS http://i67.tinypic.com/2ql6dsj.jpg

 

http://i63.tinypic.com/eve7nn.jpg

 

 

[billw55]

I am not clear on what tactic Lee discovered. Are you saying that he played a "quiet position with few tactical possibilities"? (I have not seen the game record.) In any case, it did not carry over to game 5.

Lycier, while I understand your nationalistic pride, I really don't think Ke Jie will fare any differently.

[lycier]

First three=AI

Forth=Human: Human says they won by playing unorthodox.

--------

 

Many Go players complain Human is old and weak player and should retire.

 

http://i68.tinypic.com/2yy7ez4.jpg

 

 

 

It is sure is, too many people including many Go fans become AlphaGo fans at present !

[lycier]

 

Lycier, while I understand your nationalistic pride, I really don't think Ke Jie will fare any differently.

 

Perhaps there are some language problems, what I said should have nothing to do with " nationalistic pride " .

[billw55]

Perhaps there are some language problems, what I said should have nothing to do with " nationalistic pride " .

Perhaps I misunderstood. Your are in China, you sound like a big fan of China's current top player Ke Jie, and seem to believe him decisively superior to both Lee Sedol and AlphaGo.

 

At any rate, he is a professional, and so I expect he will play AlphaGo for the right price ... which is likely to be higher than Lee's. Time will tell.

[Zelandakh]

I am not clear on what tactic Lee discovered. Are you saying that he played a "quiet position with few tactical possibilities"? (I have not seen the game record.) In any case, it did not carry over to game 5.

Having not played go since university, I could not say what the tactics were either but his comments suggest he played in an unconventional way and that was enough to throw the computer off in Game 4. For Game 5, either he was unable to reproduce the same type of position or the coding team adjusted the program to compensate. The unconventional play is the equivalent of "anti-computer tactics" in chess, which basically means creating positions that are as boring as possible. The other tactic from chess was long-term positional sacrifices such as the Benko Gambit. These methods will sometimes allow a very strong player to draw against a computer but it has been a long time since a human has won a multi-game match against a chess computer. By anaology, I would expect there to be similar kinks in the armour of AlphaGo for some time, generally in unusual positions where the usual rules are not helpful. Lee seems to have stumbled on such a weakness and I would expect the strongest Chinese players to get together now to analyse how they can best exploit that for the inevitable match that will follow.

[lycier]

Perhaps I misunderstood. Your are in China, you sound like a big fan of China's current top player Ke Jie, and seem to believe him decisively superior to both Lee Sedol and AlphaGo.

 

At any rate, he is a professional, and so I expect he will play AlphaGo for the right price ... which is likely to be higher than Lee's. Time will tell.

 

 

When we know AlphaGo Vs South Korea Lee sedol, all the people of China,Japan and South Korea are very shocked, all the peoples are very excited.

In fact, in South Korea, peoples pay close attention to this event more than to their presidential election.

In China, too many people also pay close attention to this event with over one hundred million.

For us, this is a great event,more interesting than concern with affairs of state. That's to say that a person without passion isn't real fans, just like USA NBA..

Assume AlphaBridge overcome the bridge champions team, I believe Europeans and americans also are very very excited, just like us.

So I think DeepMind founder Demis Hassabis, UK scientist, should obtain the Nobel Prize.

[lycier]

Perhaps I misunderstood. Your are in China, you sound like a big fan of China's current top player Ke Jie, and seem to believe him decisively superior to both Lee Sedol and AlphaGo.

 

At any rate, he is a professional, and so I expect he will play AlphaGo for the right price ... which is likely to be higher than Lee's. Time will tell.

 

Our professional Go champions think we maybe have a chance to overcome AlphaGo due to its weakness at present, however after 5 months to 12 months, no one have ability to overcome AlphaGo !

[billw55]

So I think DeepMind founder Demis Hassabis, UK scientist, should obtain the Nobel Prize.

Wow, you definitely have passion about this!

[lycier]

Wow, you definitely have passion about this!

 

We are most concerned about AlphaGo just passing by, then after its retirement, will leave the Go game forever.

[Fluffy]

Was the AlphaGo team allowed to adjust the program betweeen games?

[awm]

No adjustments between games. This was part of the conditions of contest.

 

Further, part of what's exciting from an AI standpoint is that AlphaGo is built on a very general machine learning framework. There isn't really human "expert knowledge" being hard coded in the way most chess programs have. This means even if they wanted to "adjust the program" between games there isn't really a way to do that.

[jogs]

What happened to the Japanese players? In the movie Go Masters the Japanese masters dominated Go in the thirties.

[lycier]

http://i68.tinypic.com/23s99xf.jpg

 

http://i66.tinypic.com/ay8x0g.jpg

 

 

AlphaGo won $1 million bonuses, lee se-dol got $170,000 appearance fees and bonuses.

AlphaGo Vs human war be impressive ,now it seems it is impossible for human to overcome AlphaGo, American technology is too strong, too terrible. AI will shape the future of human.

[barmar]

3- I think DeepMind founder Demis Hassabis, UK scientist, should obtain the Nobel Prize.

Unfortunately, there's no Nobel Prize category for Computer Science.

 

And I think Watson winning Jeopardy is an even higher achievement than this.