Life In 19x19

Without a major improvement in the AI approach to go, the computer won't be able to teach humans how to play. The current monte carlo based programs can tell you which move they think is best but the only reason they can give is that it has the best chance of winning. That is usually not a sufficient explanation of why the move is good.

For me, whether a program could defeat me is not interesting. I play go to try to understand, not just to win. When programs can "explain" moves conceptually I might be more interested. Anyhow, none of us, nor computers, can play perfect go, so for all of us it is a matter of "personal best" activity.

Anzu wrote:Chess computers are so strong, they only take one second per move to beat me. I look forward to having an ultra-fast Go engine in the future. Why play humans when your Go software moves instantly?

This may be closer than you think. The blitz version of Zen plays as a 6 dan on KGS. That is on a cluster, and blitz is slower than one second, so I don't know how strong it would be on a consumer device playing at one second per move. But it might be enough.

Anzu wrote:Why play humans when your Go software moves instantly?

1. I like my opponent taking some time, I can use that time to read, count, plan ahead*, or relax, eat a biscuit, or whatever.
2. Call me a sadist, but killing a group of a human player just seems more fun to me than killing the group of an emotionless AI. Particularly if they make some overplay and try to kill you and then you counterattack and kill them back. Revenge is sweet.
3. AIs don't make good real-life social players. Go clubs/tournaments are fun.

P.S so No.

* or review the game and scold myself for all my dumb mistakes

Uberdude wrote:

Anzu wrote:Why play humans when your Go software moves instantly?

2. Call me a sadist, but killing a group of a human player just seems more fun to me than killing the group of an emotionless AI. Particularly if they make some overplay and try to kill you and then you counterattack and kill them back. Revenge is sweet.

Oh, trust me, you get emotional when playing a computer, too.

Whenever I play my best against the chess computer and manage to win, I feel an overwhelming sense of well-being. It's just like sweet revenge.

..And even though it's just a machine, I sometimes curse and swear at it when it gives me a hard time.

CnP wrote:a game two 30-Dan AI's play.

There is likely no "30 dan" level at go.
Pro players asked how they could win against god, answered about 3-4 stones of handicap, playing black.

Even taking into account that the pro scale is more fine grained than the amateur one, that put perfect play at 9p+4 stones, so about 21p, which would be about 14 dan in the amateur scale.

Would anyway be very interesting...

If very strong AI's become widely available, I hope that the possibility of cheating does not become a significant issue. For example in the german "Go Bundesliga" the rounds are mostly played online via KGS. This is quite convenient as otherwise I would have to travel to Freiburg, Hamburg, Berlin etc. at least once every two months. In a german chess league the same model would be inthinkable. One of the benefits of Go is its friendly, noncommercial character (at least in germany) and the fact that AI cheating does not seem to play a big role, yet.

schawipp wrote:If very strong AI's become widely available, I hope that the possibility of cheating does not become a significant issue. For example in the german "Go Bundesliga" the rounds are mostly played online via KGS. This is quite convenient as otherwise I would have to travel to Freiburg, Hamburg, Berlin etc. at least once every two months. In a german chess league the same model would be inthinkable. One of the benefits of Go is its friendly, noncommercial character (at least in germany) and the fact that AI cheating does not seem to play a big role, yet.

Well since you are talking about yourself in relation to this, there already ARE programs, widely available, which even running on modest hardware, are far stronger than your current level.

We should note though, this discussion is really just about the possibilities of cheating without need for a human confederate. The model of cheating* doesn't REQUIRE a machine helper unless that "and nobody else knows" is part of the definition. Let's look at possible solutions (and the chess folks COULD maybe use the idea).

Problem: Too expensive, difficult to get the human players together at one tournament location so playing remote. But if players are each coming in remote from a private location, could be cheating (have a helper alongside, whether human or machine).

Possible solution: A number of dispersed official playing sites, thus eliminating the travel burden. These could have "proctors" seeing that cheating is not taking place. In other words, still playing remote, but not in private.

goTony wrote:Hello all,

I find GO fascinating, due to many of its different aspects. I find it amazing that AI has not defeated the top players already. And that is one of the beauties of it for me. Will any of you lose interest, or will some of the luster of GO be diminished when AI has outstripped all of us?

I would still keep playing ... its a very fascinating game. The strongest Go AI is only about 6 dan amateur, no where near defeating a top Go professional in a proper tournament. The best result it has achieved is a 20 point victory against a retired Takemiya Masaki when Takemiya gave it a 4 stone handicap in an exhibition match ...

Regards,

Cai Gengyang

I have lasted a few hours without losing interest so far.

Lets get serious here though despite the sensationalist titles of articles now swarming the internet we all know that while this is a large step for go AI it is no where near a Deep Blue incident.

We can look forward to March though where our champion Lee Sedol will take on the beast known as AI. I am suspecting a crushing victory for humanity but who knows.

There's nothing negative about Go programs getting stronger. It doesn't make humans weaker, and now we have another source to study from.

If there were to be a Go program at the level of Lee Seedol, I would be ecstatic. Can you imagine how fast you would improve playing against a professional level AI? Are the very least, you could get some bad habits hammered out of you pretty quickly.

I hope that Lee Sedol wins!

If AI is better than humans, it will be painful but I won't leave the game.

Darsey wrote:If AI is better than humans, it will be painful but I won't leave the game.

I wonder what would happen if Data played Spock. Who would win?

DrStraw wrote:

Darsey wrote:If AI is better than humans, it will be painful but I won't leave the game.

I wonder what would happen if Data played Spock. Who would win?

Data, Spock was a terrible Chess player, he could do advanced calculations in his head, tell time. But could not see a mate in a couple of moves.

https://www.youtube.com/watch?v=hEnxVwppE9M


https://www.youtube.com/watch?v=BFeMD2MfNDg

I always thought it was ironic.....