Why Deep Blue Beating Garry Kasparov Wasn't the Beginning of the End of the Human Race
How a grandmaster being taken down by a supercomputer became a turning point in how we think about (and fear) computers.
Courtesy the New York Times' Retro Report, a look back at the chess match that was about much more than chess: Garry Kasparov versus IBM's Deep Blue. In a series of matches in 1996 and 1997, Kasparov took on a Deep Blue, a supercomputer running in massive parallel, allowing it to brute-force through possible chess moves quickly enough to allow it to play chess within classic tournament time constraints. In their first meeting, Kasparov handily beat the Deep Blue, taking 4 points to Deep Blue's 2. But in 1997, after winning his first game, he became shaken by a newer, "Deeper Blue," which showed a depth and subtlety of play that the grandmaster was not expecting. When Deep Blue failed to take the bait on a trap laid by Kasparov for an easy trade, opting instead for longer positional play, Kasparov was shaken, and failed to win a single game after that.
Advertisement - Continue Reading Below
We've always been leery of the idea of machines smarter than us, but somehow the Deep Blue matches became a turning point in the mind of the media that our days at the top were numbered. Today, any off-the-shelf computer can beat even the world's greatest players, and there's plenty of media coverage about feats like IBM's Watson computer beating Jeopardy champs or Google's deep neural network AlphaGo beating grandmasters at previously unbeatable (and much more complex) games like Go. But while there are specific tasks that computers are very good at—superhuman, you might say—these feats, as IBM's Guruduth Banavar puts it, are "actually one small part of the breadth of intelligent behaviors we exhibit."
Most Popular
Deep Blue played chess in a fundamentally different way than Kasparov did, and while it may have been able to beat him, there are innumerable other things that Kasparov could do that Deep Blue never could—including having an emotional meltdown and then apologize to the press about it later that day. Computers will continue to improve, but the days of Hal shutting the pod-bay doors on us still lie far, far in the future—if they ever even happen at all. Instead, what's more likely is that machine intelligence will work alongside human intelligence, working more as helpful assistant than destroyer of the human race.
As Fei-Fei Li of Stanford puts it, "We have made a lot of progress, but it's important to understand we are closer to a washing machine than a Terminator."
For more, read Clyde Haberman's esssay over at the NYT, "Smart Robots Make Strides, but There's No Need to Flee Just Yet."
No comments:
Post a Comment