AlphaZero: What can we expect next?

Sort:
mcris

We must not fall for the "beauty" of AZ vs SF published games. Probably you folks have not witnessed many engine-engine games, there are many spectacular such games. Just an example: SF against Jonny (TCEC competition), it is on Youtube. SF played almost all game with one Rook down and went to win.

Also, they say SF had no opening book in the games. But without opening book, SF does not play the opening moves we were presented with. For example, it does not enter in QID by itself, but plays d5 early, or at least d6. Also, even after b6, it plays Ba6 and not Bb7.

breakingbad12
mcris wrote:

We must not fall for the "beauty" of AZ vs SF published games. Probably you folks have not witnessed many engine-engine games, there are many spectacular such games. Just an example: SF against Jonny (TCEC competition), it is on Youtube. SF played almost all game with one Rook down and went to win.

Also, they say SF had no opening book in the games. But without opening book, SF does not play the opening moves we were presented with. For example, it does not enter in QID by itself, but plays d5 early, or at least d6. Also, even after b6, it plays Ba6 and not Bb7.

This is true. Almost all (top) engine-engine games that don't end up in draw are beautiful in some way. AlphaZero is a major technological advance, but in terms of beautiness is overrated.

sammy_boi
Dark_Army wrote:

As most of you already know, Google's AI company Deepmind, has created a computer program that plays Chess so beautifully it's left the world's entire Chess community sitting back in awe. While the word "Genius" is always reserved for such brilliance, in this case it seems to have been relegated to an inferior option for describing just how great the artificial thinker actually is.

While it might seem like a logical choice for GM's and other great players to implement into their own games some of AlphaZero's plans for chess board domination, it's possible that humans simply do not possess the brand of intellectual depth required to follow through with those ideas. I'm sure many will try resulting in hit and miss results.

Deepmind has released only 10 of the 1300 games.

- 100 games from the starting position: (28 wins, 72 draws, 0 losses.)

- Twelve, 100 game matches from predetermined positions: (290 wins, 886 draws, 24 losses.)

The question is, why? For what reason are they holding back the other games?

We've been told that they're waiting to produce a full report on the match before releasing the other game. While that's probably true, I believe they're working on something else right now.

My speculation is that when Deepmind gets around to releasing the other games, they will also release news that AlphaZero has produced a repertoire for the white pieces (and possibly the black) that never loses. I believe they're trying to solve chess. Why not? They already did it with the ancient Chinese game of Go.

I believe however that they wont attempt to do this through a single set of moves that never looses, but rather a choice of systems where AlphaZero will always have a response to any move by the black pieces which will always lead to a white victory.

Based upon AlphaZero's performance with the white pieces, I'm fairly confident that Deepmind has pondered this possibility. It's likely they're attempting to narrow down the options into superior openings and superior lines that will eliminate the draw.

Doing the same thing with the black pieces seems like a tougher task. But with the white pieces I think it's within reach.

At any rate, I'm really looking forward to the release of the other games. I love the Chess.com analysis and commentary, especially by Daniel. 

It lost plenty of games in the 1200 game match.

And in the 100 game match it only scored 64% against a hamstrung SF.

It is not even close to a quantum leap in chess playing machines, but it is a big step for AI.

Elroch

It's not a huge leap (like its go playing brother) but most experts would say it is probably the strongest chess player.

sammy_boi

Yeah, but it's like being impressed with how good Stephen Hawking is at multiplication. Who cares how good he is at it tongue.png

(not a great analogy, but you know what I mean)

Elroch

No. I said AlphaZero was probably the strongest chess player. Stephen Hawking is unlikely to rank highly at multiplication. I would say Stockfish is probably the strongest conventional engine, but it's a bit too close at the top and depends on the version of it and its competition and the competition rules.

sammy_boi

Yes, it's a bad comparison because Hawking is probably average at multiplication. I saw some youtube video of a woman in India who could multiple something absurd, like 20 digit numbers in her head.

The point is the OP is overly impressed considering the results, and is impressed for the wrong reasons. AI offers immeasurable benefits to humanity. Being marginally better than #2 at chess is not interesting by itself.

Elroch

Doing so by teaching yourself from the rules, doing the same for go and shogi, and exceeding the knowledge achieved for each over thousands of years is both impressive and interesting.

Note that the standard of play reached in go, the most complex of the games and the one which was regarded as the least amenable to computer players for many years, was a HUGE leap above what had been achieved before. However, this is greatly assisted by the lack of draws.

sammy_boi

As I said, he was impressed for the wrong reasons, because of the potential immeasurable benefits to humanity.

sammy_boi

Imagine the AI weighting votes in a democracy, or kicking out thoroughly unfit candidates... or even eliminating human politicians completely.

What I'm saying is, imagine a world where a buffoon is relegated to reality TV and tabloid magazines wink.png 

Elroch

I too think that chess is of no importance in itself, and that there are other applications of AI which are vastly more significant.

Slow_pawn

I liked reading about alpha zero when the match results were released, and although I was impressed at the time, it faded for me. Nothing impresses me more than some of the human moves the top GMs come up with. I guess what I'm saying is that I’m way more impressed by what a mathematician can do than I am of a common, or even a sophisticated calculator. Also, to me alpha zero represents how far human diligence and intelligence has come, not how far computer programing has come.

MickeyDeadGuys
ponz111 wrote:

Am wondering if Alpha Zero or something similar could be used for investing in stocks?  If possible--some people could make a lot of noney.

There are already a lot of computerized algorithms doing trading of all kinds.  I’ve heard 75% of daily stock trading is computerized.  It’s been going on a while.  I believe the book called The Quants is about it.  Problem is during a huge market downturn like 2008, the computer trading platforms got slaughtered.  Some went bust.  Others just hit the off button, which added to the downward volatility.

ponz111
MickeyDeadGuys wrote:
ponz111 wrote:

Am wondering if Alpha Zero or something similar could be used for investing in stocks?  If possible--some people could make a lot of noney.

There are already a lot of computerized algorithms doing trading of all kinds.  I’ve heard 75% of daily stock trading is computerized.  It’s been going on a while.  I believe the book called The Quants is about it.  Problem is during a huge market downturn like 2008, the computer trading platforms got slaughtered.  Some went bust.  Others just hit the off button, which added to the downward volatility.

yes, but that was before AlphaZero and other ai research.

Godeka
sammy_boi hat geschrieben:

It lost plenty of games in the 1200 game match.

 

I don’t think many people will agree with your definition of “plenty of games”. AlphaZero lost 5 games with white and 19 games with black. That are 24 games of 1200.

mcris
Godeka wrote:
sammy_boi hat geschrieben:

It lost plenty of games in the 1200 game match.

 

I don’t think many people will agree with your definition of “plenty of games”. AlphaZero lost 5 games with white and 19 games with black. That are 24 games of 1200.

He is American, you are German. I trust his English better.

fewlio
mcris wrote:
Godeka wrote:
sammy_boi hat geschrieben:

It lost plenty of games in the 1200 game match.

 

I don’t think many people will agree with your definition of “plenty of games”. AlphaZero lost 5 games with white and 19 games with black. That are 24 games of 1200.

He is American, you are German. I trust his English better.

 

Just want to say, english is a german language derived from the angles and saxons of northern germany, however some genius ditched the masculine, feminine and neuter articles, which plague latin based, slavic and even german for some reason!  What a stupid idea.  This is why I laugh when people say english is hard...no it's not nearly as hard as those with so many cases (for adressing different people of different sexes and status in life) and gender articles.  This tree is feminine, this telephone is masculine, utter nonsense, and it makes it so you have to learn not just how to spell a word, but also its "gender status."  A total waste of effort and brain space.

jbent02

"

My speculation is that when Deepmind gets around to releasing the other games, they will also release news that AlphaZero has produced a repertoire for the white pieces (and possibly the black) that never loses. I believe they're trying to solve chess. Why not? They already did it with the ancient Chinese game of Go.

I believe however that they wont attempt to do this through a single set of moves that never looses, but rather a choice of systems where AlphaZero will always have a response to any move by the black pieces which will always lead to a white victory." PLEASE NO, DON"T EVER SOLVE CHESS.

Godeka
mcris hat geschrieben:

He is American, you are German. I trust his English better.

 

For sure. If you think “plenty of” means something different than “viele“ you can correct me.

 

@fewlio:

> however some genius ditched the masculine, feminine and neuter articles

grin.png

gented
Godeka wrote:
breakingbad12 hat geschrieben:

Something seems fishy about all that indeed. They don't disclose the games, they don't disclose their engine for public use, their engine doesn't appear in the engine ranking, even tho it has potential to be in first. But most importantly, why did they take so long to make a chess engine if they already made a go engine? If deep mind is truly neural and if they only need the rules of the game and the final result as input (after self training), why don't they apply that to a variety of different games?

 

They already disclose a lot: the basic concept and NN structure, games, multiple statistics. I lasted two years to develop AlphaGo Fun (end of year 2015) to AlphaGo Zero (October 2017). If I remember correctly there were three papers, two events (matches against Lee Sedol and Ke Jie), there was a variant of AlphaGo Master on a public Go server (which won 60 of 60 games against professional players on Tygem), and the AlphaGo Teach tool was released. In total 213 games played by all AlphaGo versions from 2015 to 2017 are available.

 

What is missing are the latest games of AlphaZero. Go and Shogi games are completely missing, for chess at least 90 games are missing (or 1210 if you count the games started with fixed openings).

 

I hope DeepMind will release the 90 chess games. I don't see any reason why they are hold back. There must be a reason of course, maybe DeepMind simply does not care about interested chess players. This can be disliked and criticised, but there is nothing fishy about it. They released a lot of data before, and AlphaZero is clearly a derivation of previous works. That a completely self learning AlphaGo-like NN works is proofed.

 

You must be kidding, right? They did not disclose anything, except for random buzzwords left and right in a very poor paper. Can you point us, otherwise, to the right reference where they describe the way the calculate and assign probabilities to moves? And don't throw me the MonteCarlo Tree Search buzzword, because that alone means little to nothing (and one can build at least 100 different algorithm using MCTS performing in several different ways, eventually).