picture

Go Wizardry

All About the Many Aspects of Go
We have millions of friends around the world... and they all play go!

State of the Art in Artificial Intelligence

image

Anyone who is interested in the game of go immediately takes notice when dramatic events in regard to the game take place. In the last year or so, nothing has grabbed the attention of go players throughout the world like the appearance of AlphaGo on the scene. At first, it was surprising when a very strong amateur player lost an informal match to the program, but then the "world champion," Lee Sedol (who is "world champion" only because he has won a number of lightning go championships giving him that designation by the sponsors) lost another informal match and finally a semi-formal event was organized pitting the program against the Chinese champion, Ke Jie which AlphaGo won. After that, AlphaGo won against other players in lightning go games online, including against Iyama Yuta, holder of all of the major titles in Japan.

Regular visitors to GoWizardry.com know that I am not impressed with AlphaGo. I think that the games have been rigged. But my opinion is irrelevant. The fact that a computer program has defeated top professional players, regardless of the circumstances, is noteworthy. And therefore I have written a couple of postings here expressing my thoughts. I am doing the same once again.

A couple of recent developments have spurred my decision. First, the Deep Mind team that created AlphaGo has made much of its research available and that should be noted publicly as well as promoted to all with an interest in the development of Artificial Intelligence. In accordance with that imperative, I pass along the following message that I received the other day:

Alpha Teaching Tool has been released. 230,000 pro and strong online 9d amateur games and 75 AlphaGo games reviewed. 6,000 opening sequences were selected (if used six times). AlphaGo Master evaluates 370,000 positions for black’s win rate. https://alphagoteach.deepmind.com/

What does this mean to me, personally? Nothing. Not that I disparage this action; on the contrary, I applaud it. But it has absolutely no effect on me as a go player, a writer of go books or a commentator of events on the go scene. It is irrelevant as far as I am concerned.

But more than that, what does it mean it terms of the theory of the game of go? Nothing. Not only does it mean nothing, it never will.

Let me explain. The late Takagawa Kaku, who won nine consecutive Honinbo titles and is considered one of the greatest players in go history, once wrote an essay in which he had the gods of go sitting before each other to play a game. The first god lays a stone on the board and the other, reading out the game to the end resigns! "I see that I cannot win," he declares.

In another famous article, Kajiwara Takeo 9 dan, one of the major go theorists of 1960s and 70s who developed several joseki and other important strategic principles, famously declared after the second move had been played on the board, that "the game is over."

What is one to make of this?!

It is all confusing, but a recently published book by the former Chess World Champion, Garry Kasparov, offers fascinating speculations about the subject of game theory, analysis under strict tournament conditions and artificial intelligence. He puts everything in the context of his own matches against computer chess programs, including the famous one against Deep Blue. The result is an absolutely amazing account of a landmark event in the history of both chess and artificial intelligence.

image

Deep Thinking

Where Machine Intelligence Ends and Human Creativity Begins

Garry Kasparov, Former World Chess Champion

Copyright 2017

PublicAffairs, 1290 Avenue of the Americas, New York, NY 10104

ISBN 978-1-61039-786-5

Hachette Speakers Bureau (hachettespeakersbureau.com) 866-376-6591

image

Here is an excerpt from the book:

Chess computers don’t have psychological faults, but they do have very distinct strengths and weaknesses, far more distinct than any equivalently strong human player would have. Today, they are so strong that most of their vulnerabilities have been steamrolled into irrelevancy by the sheer speed and depth of brute force search. They cannot play strategically, but they are too accurate tactically for a human to exploit those subtle weaknesses decisively A tennis player with a 250 m.p.h. serve doesn’t have to worry too much about having a weak backhand.

That was far from the case back in 1985. Tactical calculations were still a computer strength, but only shallow sequences three or four moves deep. This was more than enough to beat most amateurs consistently, although strong players became adept at setting tactical traps that were too deep for computers to see. It seemed paradoxical that the machine’s strength of flawless calculation was also a major weakness. The brute force "exhaustive search" method of checking every one of millions of positions also meant that the search tree couldn’t reach very deep. If you could find a tactical threat that struck the decisive blow four moves (eight ply) away when the computer could only see three moves (six ply) deep, it wouldn’t see it coming until it was too late. We call this the "horizon effect," exploiting that the machine can’t see beyond its search "horizon."

It is clear that Kasparov is has spent a lot of time pondering questions about AI in general and his experiences in dealing with computers over the board specifically. This is natural, since he has been asked about the matches dozens, if not hundreds of times. But also, his chess career could not have continued indefinitely, and he now makes a living giving lectures about strategies in business and administration to corporations and institutions all over the world. He lives in Manhattan.

Besides discussing these matters, in "Deep Thinking" Kasparov displays a remarkable command of a wide range of subjects, including both scholarly journals and popular culture. And as an author myself, I carefully check every sentence for mistakes. I have noticed that the Wall Street Journal, which used to be meticulous in avoiding any misprints, has let some creep into its pages. "Deep Thinking" has none, although there was one mistaken repetition of a phrase. Impressive.

However, no matter how meticulous and well-read Kasparov is, he can also display a lack of historical knowledge, as in the following paragraph.

Emotional influence is only one of many the ways in which humans act irrationally and unpredictably. Economic theory is predicated on the fact that people are "rational actors," that we will always decide based on what is in our best interests. This is probably why economics is called the "dismal science" and why there is a saying that economists have as much effect on the economy as weather forecasters have on the weather. Humans often aren’t rational at all, not in groups and not individually.

Perhaps it is my background in economics (since I concentrate on economics, business and finance in my work as a professional Japanese interpreter/translator) that led me to discover years ago that economics is called the "dismal science" because of the work of Thomas Malthus, who predicted dire consequences (still dubbed "Malthusian") when a population explosion outstripped food production. (Incidentally, it may be reasonably argued that Malthus was partially correct since upwards of ten million children die every year worldwide due mainly to malnutrition and disease.)

Kasparov ends with all sorts of conclusions about the subject of artificial intelligence, far too many to relate here. But the following excerpt will give a sense of his thoughts.

Does it really matter what is or isn’t "intelligent" by some definition, no matter how well argued? I concede that the more I learn about it, the less I care. Chess is the perfect example of Larry Tesler’s "AI effect," which says that "intelligence is whatever machines haven’t done yet." As soon as we figure out a way to get a computer to do something intelligent, like play world championship chess, we decide it’s not truly intelligent. Others have pointed out that whenever something becomes practical and common, it stops being called AI at all. It’s another illustration that these narratives only matter for a brief point in time.

As a chess player, I found the detailed discussion of the 1997 match between Kasparov and Deep Blue to be fascinating. I watched it with interest at the time, but despite seeking out as much expert analysis that I could, that did not come close to what Kasparov has to say.

The match was for six games. Kasparov won the first game, and that gave him a false sense of confidence. Then, inexplicably, he resigned the second game in a position where a draw was right before his eyes! When his own analysis team pointed that out, Kasparov was dumbfounded. It shook his confidence to the core. And that in turn affected his play to the end.

Besides that, the sponsor of the match, IBM, does not come across as acting ethically. It employed Russian-speaking security guards, presumably to spy on Kasparov and his team, although Kasparov does not come right out and say that. And in the final game of the match, when the team analyzed a variation of the Caro-Kann Defense, an opening that Kasparov had never played in his life, the possibility of a knight sacrifice was discussed. However, the consequences of such a sacrifice are so speculative that computers at that time would never make it. Later, one of the Deep Blue grandmaster consultants publicly acknowledged that the team had programmed Deep Blue specifically to make that sacrifice! The night before the game! Very suspicious.

This is a literate, highly enjoyable book to read. I recommend it without reservations.

Those who wish to comment on the opinions expressed here may send their thoughts to info@GoWizardry.com. The most interesting responses will be addressed in future postings.

Robert J. Terry

Tagged as: ,

Leave a Reply

Books

book cover

Go on the Go Collection: Volume I

Three booklets have been assembled into the collection here.

Buy this Book at Amazon

Go For Everyone

Go For Everyone

A New Method for Learning to Play the Game of Go

Buy this book

Book Cover

Journey to the West

This is a semi-autobiographical novel that depicts a unique American success story; a rags to riches tale of a man escaping his humble origins to make millions of dollars, but then he throws it all away due to the ancient character flaw of hubris.

Buy this Book at Amazon