Human vs. Machine: Kasparov’s Legacy
“It’s a machine. At the end of the day, it’s stupid.” Garry Kasparov, 1997
Do you ever worry that you could be replaced by a machine? With ChatGPT and other tools on the rise, concerns like these seem more relevant than ever.
They are not entirely new, however. Around 30 years ago, the world was already on a quest to find out whether machines could outperform humans intellectually. One of the most prominent duels of “human vs. machine” was the chess match between Garry Kasparov and IBM’s supercomputer Deep Blue.
At the beginning of the challenge, in 1996, Kasparov didn’t seem to be worried that a machine could take his place as the best chess player in the world. Soon, however, he had to change his mind.
In this article, we will take a look at Kasparov’s journey of discovering, defying and ultimately collaborating (in a legal way!) with computer chess programs. The two matches against Deep Blue in 1996 and 1997 are central in this story, but let’s first take a step back.
Kasparov’s interest in computers
Kasparov always searched for new ways to improve his chess. Computers seemed like a huge opportunity. He got his first one after the 1983 Candidates semi-final match against Korchnoi. That was one of the first computers in the Soviet Union!
In 1985, shortly before becoming World Champion, Kasparov met Frederic Friedel, a fellow computer enthusiast. Together they brainstormed how computers could simplify chess tournament preparation. Back in the day, chess players had to carry around huge piles of paper with opening analyses and tournament games. Could a computer do that instead?
The result was the very first digital chess database. Kasparov received his copy of it via mail and started using it for tournament preparation in 1986.
Could computers outplay humans back then?
If even databases were a new thing back then, it is unsurprising that chess engines were not nearly as good as they are today. They were, however, able to beat amateur players, and in short time controls even titled players. But Kasparov would reliably outperform machines. In a simultaneous exhibition in Hamburg in 1985, he played against 32 of the world’s best chess machines and won against all of them.
It was only ten years later that a computer by the name of “Chess Genius” managed to defeat Kasparov. This was in a rapid time control game, though. The real challenge for chess programs was to outperform masters in classical time controls. While computers had an advantage in short tactics, humans still had a deeper strategical understanding of the game, which plays a bigger role in long time controls. A computer beating Kasparov in classical chess had yet to be developed.
The development of Deep Blue
In 1988, doctoral student Feng-hsiung Hsu developed an advanced chess machine called “Deep Thought”, first on his own and then as a part of the IBM team. Deep Thought was the first computer to win a major tournament against grandmasters in 1988. However, it was still not strong enough to beat Kasparov when they faced each other in a two-game match in 1989.
That only motivated IBM to continue improving their system. The next version of Hsu’s machine was called “Deep Blue”. A team of young engineers was working on the ambitious goal to make the machine powerful enough to beat Kasparov. They got help from US Chess Champion Joel Benjamin for the opening preparation, with a specific focus on Kasparov’s repertoire.
The result of their work was a 1.4-ton machine that was ready to compete with the world’s best player.
The Matches
The matches between Garry Kasparov and Deep Blue were not just about chess. It was a duel between a human and a machine, and the whole world was watching to see where humanity stood in this battle. One can only imagine the pressure Kasparov must have been under with headlines like “The Brain’s Last Stand”.
1996 Match
The first match against IBM’s Deep Blue was held in February 1996 in Philadelphia. There was a lot of prize money in it: $400,000 for the win and $100,000 for the loss. The first game already yielded a sensational result: Kasparov lost! That was the first time in history that a computer program beat the world champion in a classical game.
In an interview after the game, Kasparov said that he didn’t really know what to expect from the machine. Without access to any of Deep Blue’s previous games, he couldn’t prepare properly. It was only a few rounds into the match that he could devise a strategy. His anti-computer approach focused on positional play, keeping the game closed with as many pieces on the board as possible.
Kasparov went on to win this match 4–2, proving that his confidence in his superiority over the machine was justified.
1997 Match
Shortly after the first match, Kasparov and IBM agreed to play a rematch in May 1997 in New York. This match would be the one to change history, and like most historical events, it wasn’t free from controversy.
The stakes for this match were even higher than in the first one: IBM had not only increased the prize money to $700,000, but had also invested heavily into Deep Blue’s improvement in the intervening year. The latter resulted in the machine growing both in size and in computational strength: Deep Blue could now calculate up to 200 million positions per second!
Kasparov didn’t have any information on what exactly was changed in the machine’s programming, so once again he had little grounds to prepare for the opponent. He was able to win the first game, though, after Deep Blue made a mistake on move 44 which is said to have been caused by a “bug” in Deep Blue’s program.
The second game was where the controversy started; to be exact: on move 37 when Deep Blue opted for a more positional, solidifying move when it had the chance to grab a free pawn with Qb6 instead.
Kasparov couldn’t believe that a machine was able to play like that. It led him to suspect that IBM was receiving information from a strong grandmaster during the game. The tense atmosphere of the match only added to the suspicion: IBM had previously refused to send Kasparov information on the machine’s programming, and now the machine was kept in a separate room that was watched over by guards; besides that, frequent problems with the computer made it necessary to reboot it, even during games.
After the fatal second game, Kasparov couldn’t really recover. The man-machine duel continued with three draws and the last round was won by Deep Blue after a completely unexpected piece sacrifice on move 8.
To see an analysis of all games from the Deep Blue matches, check out this article.
The defeat was a shock to Kasparov. It was the first chess match in his whole career that didn’t end with a victory for him.
The aftermath
The loss didn’t deter Kasparov from continuing his quest of discovering new chess technology. In a later interview he said that after the Deep Blue matches he realized that rather than competing against the machines, humans should work with them.
In 1998 Kasparov pioneered a new chess discipline aiming at bringing together human skills and machine intelligence: a form of chess called “Advanced Chess” where each player collaborates with an engine. Despite this new approach to computer chess, he still went on to challenge newer computer programs in 2003, namely Deep Junior and Fritz. Both of these matches ended in a tie.
The topic of chess and technology never stopped fascinating Kasparov. He talked about his experiences and his perspective on technology in public and also wrote a book about the Deep Blue matches called “Deep Thinking” (2017).
Conclusion
Deep Blue’s historic victory over Kasparov didn’t mean that chess would become a useless occupation for humans. Quite the opposite: even though humans don’t stand a chance against chess engines today, the game is more popular than ever. What this shows us is that the rise of technology in a field doesn’t mean that human activity in it dies out. So if you’re scared that ChatGPT will replace you, keep that in mind.
In his TED Talk from 2017 with the title “Don’t fear machines. Work with them!” Kasparov highlights the fact that, while machines are superior to us in some areas, humans still prevail in others, and it is this combination of skills we should focus on. Let’s conclude this article with Kasparovian words that show a more nuanced and balanced take on technology than his above quote from twenty years before:
“Machines have calculations, we have understanding. Machines have instructions, we have purpose. Machines have objectivity, we have passion.” Garry Kasparov, 2017
