PCB Assembly Services - Screaming Circuits: Kasparov vs. Deep Blue, Feb 10, 1996


Kasparov vs. Deep Blue, Feb 10, 1996

A Few Words from Sara Shepherd, Screaming Circuits Contributing Author...

What the famous chess match continues to teach us about the limits of AI and the richness of our own humanity

Twenty four years ago, chess grandmaster Garry Kasparov traveled to New York City and pushed his way through a crowd of reporters and flashing cameras to take his place at the chessboard. 

Though he had easily beaten this opponent a year before, he did not appear entirely confident. The air was charged with excitement and tension, and one could hardly blame Kasparov for feeling a great amount of pressure and responsibility. After all, countless news organizations around the world had said he was playing for the future of humanity itself.

Kasparov turned to face his opponent and was met with a blinking monitor display. Deep Blue, IBM’s chess-playing supercomputer, was booted up and ready for a rematch.

The now famous 1997 six-game chess match of man vs. machine, played over a course of nine days, captivated the world. Never before had a chess match been so widely covered by the media or so romanticized by the spectators. 

Kasparov was regarded as the greatest living chess player at the time, and one of the best to have ever played the game. He had handily defeated the Deep Blue a year earlier in 1996 (though the computer had won a single game in the six-game match). 

The team behind Deep Blue had been developing successive generations of chess systems since 1985. The project began as ChipTest, which used a special VLSI-technology move generator chip developed by Feng-hsiung Hsu, Thomas Anantharaman and Murray Campbell at Carnegie Mellon University. (VLSI stands for Very large-scale integration, the process of creating an integrated circuit by combining millions of MOS transistors onto a single chip.)

After graduating, the ChipTest team was hired by IBM Research to continue their project, with the goal of creating a chess machine that could defeat a world champion. The project was briefly renamed Deep Thought after the computer in The Hitchhiker's Guide to the Galaxy — a fictional computer which spent 7.5 million years to determine that 42 was “the Answer to The Ultimate Question of Life, the Universe, and Everything.”

In 1989, Deep Thought became Deep Blue, a play on IBM's nickname, "Big Blue." This was the fifth generation of the  project and utilized a hybrid hardware approach combining normal processors and chess accelerator chips. The normal chess moves and planning were handled by the processors, while the complex tasks were computed by the chess accelerator chips. (For the hardware nerds out there, Deep Blue had 30 nodes with 480 special-purpose VLSI chess accelerator chips per each node and 120 Mhz P2SC microprocessor per each node.)

Before the rematch, IBM’s website bragged that Deep Blue could calculate 200 million board positions per second. “Incidentally, Garry Kasparov can evaluate approximately three positions per second,” it added snidely. 

The 1997 rematch took place over the course of 9 days in May of 1997. Kasparov won the first game, lost the second, and then drew the following three. The final game lasted only an hour. Deep Blue traded its bishop and rook for Kasparov’s queen, after sacrificing a knight. Kasparov resigned after Deep Blue’s 19th move, the first and only time in his career he had ever conceded defeat. 

Deep Blue’s victory was hailed as proof that computing technology had finally reached the level of human thinking. The stories of science fiction were rapidly coming true - humans and their paltry brains would be left in the dust as machines took over. 

But was it really the superior playing of Deep Blue that had defeated Kasparov?

“Everybody was surprised that Kasparov resigned because it didn’t seem lost,” Grandmaster John Fedorowicz said of the game in an interview. “We’ve all played this position before. It’s a known position.” 

At the time, when asked about his decision to resign, Kasparov told reporters “I lost my fighting spirit.”

Later, careful analysis of the games and Deep Blue’s computing logs revealed several surprising points. The first is that Deep Blue made many errors. The second, and possibly most important, was that a glitch in the first game caused Deep Blue to make a random move, instead of one his programming should have chosen. Kasparov misinterpreted the glitch as a deeper strategy, and the seed of fear and doubt was planted in his mind that this machine was far more powerful than it actually was. 

In the documentary Game Over: Kasparov and the Machine, Kasparov recounts his fears of playing against the “terrible, faceless monster” and that playing against Deep Blue was “like a wall coming at you.” Indeed, instead of using sophisticated AI, Deep Blue mainly relied on brute computational force to evaluate millions of positions and pressure its opponent in a tiring, defensive position.

Fearful of Deep Blue’s capabilities, Kasparov abandoned his usual initiative, swashbuckling style, and took up a defensive, wait-and-see approach. In his book Deep Thinking, Kasparov says this played in the computer’s favor and was the reason for his defeat.

So perhaps Kasparov lost not to Deep Blue but to himself, his own fear and self doubt. Deep Blue was a superior chess player, but benefited from never becoming tired or emotional. Of course this story - humans losing to a machine because of their humanity - was not as exciting to the public.

Chess has come back into popular culture today thanks in large part to the Netflix series The Queen’s Gambit, which follows the fictional chess career of Beth Harmon as she battles addiction to become a grandmaster. As the story unfolds, it becomes clear that Harmon is defeated not by superior players but rather by her own doubts and alcoholic binges. By trying to numb her fears and anxiety with drugs, she loses what makes her an exceptional player: her access to intuition and creativity.

Chess may be thought of as a logical game, but Kasparaov’s defeat shows us that it is also a hugely psychological and emotional one. A player’s ability to withstand small defeats and maneuver into a winning position are what make the game so exciting. It is creativity, stamina, and perseverance that make it a joy to watch.

Kasparov has since gone on to write and speak about the possibilities when human ingenuity is paired with the tools of AI. His 2017 TED Talk “Don't Fear Intelligent Machines. Work With Them” revisits his loss to Deep Blue and the limits of computerized thought. The ability for Deep Blue to calculate 200 million board positions per second is meaningless if the program cannot apply those calculations creatively towards an end goal. 

AI has advanced greatly since 1997. More recently, a computer program named AlphaGo was developed to master the game of Go. It pairs a search tree similar to Deep Blue’s to analyze possible moves with a complex neural network that analyzes the game’s outcome based on thousands of previous plays (the program plays against itself, amassing thousands of years of human experience in only a few days). 

In October 2015, AlphaGo played its first match against the reigning three-time European Champion, Mr Fan Hui and won the first ever game against a Go professional with a score of 5-0.

“I thought AlphaGo was based on probability calculation and that it was merely a machine,” said Lee Sedol, winner of 18 World Go titles. “But when I saw this move, I changed my mind. Surely, AlphaGo is creative.”

Attempts to create thinking machines force us to examine what intelligence really is. The deeper we dig, the more we discover the depths of human thinking lie not only in the brain's ability to reason, but also in the feelings of the heart and the creativity of the soul. 

Despite a computer’s supposed superiority in the game, chess is more popular than ever, with stores reporting 200% increases in the sales of chess boards and timers. Our desire to play, create, and solve does not appear at all dampened by modern technology, and that is a good thing.

[Editor’s note:  if the simulation is so effective, such that the human in opposition cannot tell that a machine is not operating as a creative, emotional entity; cannot tell that the machine is not sentient, does it really matter whether it is or not?]


Post a comment

If you have a TypeKey or TypePad account, please Sign In.

« Component supply - Is it panic time? | Main | Nationwide Weather Delays - and, yes we can quote your PCBs »