Computers beating top human players -- good or bad?

[RobertJasiek]

The very best way to improve is to find a really strong teacher [...] who can adjust his/her play to the rank level of the student and is familiar with the mistakes commonly made by players at various levels. Such a teacher would play even games with the student, posing "problems" for the student at the appropriate level by creating situations in the game that require the student to practice specific technique(s). Some of this would be through deliberately making mistakes for the student to recognise and punish. And such a teacher can give reasons for moves such as "slow", "inefficient shape", "need to fight here", "inconsistent strategy", etc. With this sort of teaching the student will be far less likely to learn bad technique which would have to be unlearned later. Unlearning takes more time than learning correctly in the first place.

This is pretty much correct with these changes:

1) The teacher must be able to identify all the important mistakes of the pupil (and not just the 30% fitting the teacher's restricted own vision) and give him related general advice.

2) The best learning is combination of lessons, playing, reviewing one's own games, go theory literature, problem solving to train especially reading and endgame and studying strong players' games.

[topazg]

This is pretty much correct with these changes:

1) The teacher must be able to identify all the important mistakes of the pupil (and not just the 30% fitting the teacher's restricted own vision) and give him related general advice.

I think there is no teacher (or any human being) that has unrestricted vision, and restrictions in ability/understanding are equally important (and equally restricted).

[RobertJasiek]

Possible, but still teachers' abilities to identify major mistakes vary extremely from teacher to teacher.

[Garuseimasahi]

Go atracts me because I wish to understand the game. A computer will not change that, one way or another. (Though winning is nice, too.)

[lobotommy]

The last big jump in performance of the programs was when they went from AIs to using MCTS. That was a major breakthrough and one that could not have been predicted in advance. It would not surprise me if fine tuning of MCTS programs might get them 1-2 stones stronger than they are now. It would also not surprise me if it turned out that we have already reached the limit of performance for the MCTS algorithm and that further advance will require a new breakthrough.

When MCTS programs were introduced, they were mid-SDK, only a few stones stronger than the strongest existing bots. They have since more than doubled that gap. Zen became the first KGS 1 dan in 2009. Now we have KGS 6d bots, a tremendous improvement in just three years. And the huge jump from mid-SDK to 6d can't be explained by increased processing power alone. It just doesn't seem like go AI development has stagnated since the introduction of MCTS. Even back when Zen made 1d, people would just laugh if you told them we'd have KGS 6d bots in just three years. With this kind of history, it's hard to be pessimistic about the future of go AI.

Zen bot is 6Dan on KGS? Take a look at my go strenght and compare. So, I am(was?) 4kyu in EGF, I'm playing close to 2d on tygem/kgs/wbaduk. So if 4kyu get 2d with not too much problem on internet servers and Zen is only 4 stones stronger on KGS than 4kyu player it simply means to me that it is probably no more than european 2d. 3dan at best. And his high rank was achieved mostly because of blitz games. I think it is still far far away from even mid dan section. Maybe stronger players could let us know if they were beaten by this bot?

[palapiku]

Zen bot is 6Dan on KGS? Take a look at my go strenght and compare. So, I am(was?) 4kyu in EGF, I'm playing close to 2d on tygem/kgs/wbaduk. So if 4kyu get 2d with not too much problem on internet servers and Zen is only 4 stones stronger on KGS than 4kyu player it simply means to me that it is probably no more than european 2d. 3dan at best. And his high rank was achieved mostly because of blitz games. I think it is still far far away from even mid dan section. Maybe stronger players could let us know if they were beaten by this bot?

My post was about the rapid progress of Zen - from KGS 1d to KGS 6d. Regardless of any absolute strength of KGS ranks, that's a five-stone improvement of the kind that most humans take far longer to achieve.

[Mike Novack]

Relative strength is all we have data for.

a) Not a lot of data how much weaker (relative to humans) when playing at slower time controls. Remember, the program would actually be playing stronger at these speeds, just not helped as much by the additional time as the human. A stern chase is a long chase.

Perhaps the best we can do is see if we can find data for the program running at the same (fast) time control but on different hardware.

b) There have been some matches played between humans of (presumably) known strength and computer programs. Again mainly useful for relative strength comparisons -- the bots do play each other fairly frequently.