The main idea of this book is to review author’s experience as the first chess champion loosing to AI and using this experience to discuss past, present, and future development of AI and consequences of this development. The main objective here for author is to find reasons to be optimistic and demonstrate that advance of computers does not mean decline or even obsolesce of humanity.
ONE The Brain Game
Here author discusses the nature of chess as the brain game and commonly recognized link between ability to play chess and intelligence. Correspondingly this game was often used in all discussion of AI as a benchmark of its abilities. After that author moves to his experiences as chess prodigy in the Soviet Union where this game was treated as an important factor in competition with Western world for superiority in the area of intellectual development. Author briefly reviews history of chess related Cold war episodes and his role in them when he became world champion in 1985 just before Soviet Empire started its rapid move to demise.
TWO Rise of the Chess Machines
This chapter briefly reviews the story of development of chess related software and hardware from the earlier attempts when computers just started in late 1940s to Deep Thought specialized computer that was able successfully play at the Grandmaster level.
THREE Human versus Machine
Here author describes how computers were introduced into chess world and what kinds of strategic algorithms were used from initial stress on material weight to brute force positions crunching. He also discusses his own participation in the development of chess computers. The pinnacle of initial achievement was the program called Chess that was capable to win in a week human tournament.
FOUR What Matters to a Machine?
This is more about strategy and tactics used in development of AI algorithms and use of Chess for their development where the clear rules of game and complete information defined by rules provided a type of experimental tool similar to use of drosophila in biology. It is also about deficiencies of such approach because the main assumption about working of mind similarity to a computer proved to be incorrect.
FIVE What Makes a Mind?
Here author moves to discussion of properties of human mind and psychological aspects of chess. In process he discusses popular, but not necessarily well-founded idea about 10000 hours required for mastering any skill. Author claims, based on this own experience both as progeny and later as teacher, that initial genetic makeup of individual could not be discarded and 10K hours would produce different results in different people. In the second part of the chapter author discusses various anti-computer strategies that he developed to overcome computer’s advantage in tactical calculations and lacking of psychological component that could be debilitating for human player.
SIX Into the Arena
Here author briefly returns to his experience playing for championship against Karpov – KGB preferred winner, then moves to discussing DARPA initial development of AI. This follows by author’s discussion of complexity of rules in culture and lives, which is dramatically different and higher than simplicity of rules in chess. As example he looks at machine translation of human languages. The last part of the chapter is the story of author’s match against Deep Thought computer in October of 1989, which author won, but it was already a serious game.
SEVEN The Deep End
This chapter is the first part of the culmination of the story because it is about the match with Deep Blue. Author still won this match, but it was not that easy any more and Deep Blue demonstrated that it is getting close. Author also briefly describes chess revolution that occurred with dramatic improvement of PC that obtained enough computer power to provide vast majority of players with a partner at the level they want to have, even if it is pretty high level.
EIGHT Deeper Blue
This starts with discussion of improvement in computer power that practically eliminated need for super complex algorithms by bringing to the table tremendous computer power. It follows by the story of the second match with Deep Blue, but not before author reviews meaning and advantages and disadvantages of rematches. At the end author discusses prelude to the second game and his overestimation of time needed to Deep Blue team to fix problems and achieve significant improvement in its game. He also points out that IBM hired quite a few Grandmasters to work with Deep Blue.
NINE The Board Is in Flames! TEN The Holy Grail
These two chapters represent very detailed game-by-game and nearly move-by-move description of the second match when human chess champion (author) actually lost to the computer. It also includes discussion of several related controversies and author’s psychological ups and downs that had significant impact on the quality of the game.
ELEVEN Human Plus Machine
The final chapter is about combining AI and human intelligence in teams, which allows for better performance than either of them separately. Author discusses quite a few areas like GPS where AI added tremendously to the quality of life and our ability to achieve results. This obviously includes chess where computer programs dramatically improved training opportunities for players. One very interesting point here is that young players now learn quite differently from the old school that author went through. They are playing a lot more and consequently obtain a lot more practical skills, while learning a lot less theory. Then author discusses AI impact on innovation. Specifically in chess Grandmasters now have opportunities to run any innovation with AI before using it in high stakes game, allowing them to test much more approaches than was possible before. At the end of chapter author discusses more details of human vs. AI strengths and weaknesses and how to combine them to achieve maximum effect.
Conclusion: Onward and Upward
The final conclusion is pretty much that future is bright. The author believes that while humans good at developing machines to do their work the work itself will not disappear, but rather expand providing more and more interesting things to achieve and tasks to complete because humans have purpose and AI does not.
MY TAKE ON IT:
It is very interesting recount of the human vs. AI struggle in probably one of the simplest computer friendly applications that become proving ground for initial AI development. I am mainly in agreement with author that it is not a threat to humanity as long as humans would not decide to make it a threat either from military or some other considerations. Actually I am believer that humans are to the high extent are product of their experiences, so even if we create AI and feed into it huge amount of information and allow it to get skills via experience like self-teaching AI program, it would not be substitution for humanity, but rather just another tool, granted a very sophisticated one. The key feature that makes humans what they are, in my opinion, is their self-directed character that provide for ability to obtain idiosyncratic experience and form unique personality. I would not exclude possibility of AI computer that is designed to do just that: to learn from “parents”, obtain unpredictable experiences, direct its own development in some direction not necessarily controlled from outside, and eventually develop its own self and personality. I am pretty sure that it will occur in some kind of experimental form. However I do not think that it would go anywhere beyond experiment because eventually if it is successful all that we going to get is another human being with silicon rather than biological basis, but human being nevertheless. And since we have much easier and cheaper ways to produce those and all advantages that such AI person would have from more memory or higher speed of calculations could be easily matched by non-personified AI acting under human direction, it will probably never go beyond experimental stage.