Google's AlphaGo Uses AI To Beat World Champ 'Go' Player
Nearly 20 years after IBM computer Deep Blue first beat reigning world chess champion Garry Kasparov in a six-game match, a new generation of machine intelligence -- Google's AlphaGo -- has bested professional Go player Lee Sedol in the first game of a planned five-round match.
Go has been considered a much harder game for a computer to beat than chess, and it's played on a much larger grid -- 19 x 19 versus 8 x 8 -- with many more possible moves. Sedol, who's from South Korea, has the second-highest number of international titles in the game today. He resigned in his first match against AlphaGo at 10 a.m. local time in Seoul, where the competition is taking place.
Google's AlphaGo challenge against Sedol (pictured above) involves five games to be played between today and Tuesday. The contenders are vying for $1 million in prize money, which Google's DeepMind -- AlphaGo's developer -- plans to donate to UNICEF, STEM and Go societies if it wins.
AlphaGo 'Did Not Back Down'
"They were neck-and-neck for its entirety, in a game filled with complex fighting," DeepMind founder and CEO Demis Hassabis wrote today on Google's Asia Pacific blog. "Lee Sedol made very aggressive moves but AlphaGo did not back down from the fights. AlphaGo took almost all of its time compared to Lee Sedol who had almost 30 minutes left on the clock."
First played in China more than 2,500 years ago, Go had until recently proved too complex for smart machines playing humans. However, in October, DeepMind's AlphaGo won all five games against professional Go player Fan Hui -- a first for game-playing artificial intelligence.
"Go is a profoundly complex game, which is why this has always been regarded as the outstanding grand challenge for artificial intelligence," Hassabis noted in a blog post yesterday.
Google isn't the only tech company hoping to build an artificial Go champion. Facebook is also tackling the challenge with a system it's developed called Darkforest. Facebook founder and CEO Mark Zuckerberg said in an update in late January, "Scientists have been trying to teach computers to win at Go for 20 years. We're getting close, and in the past six months we've built an AI that can make moves in as fast as 0.1 seconds and still be as good as previous systems that took years to build."
No AI 'Singularity' Moment Yet
"This is the Deep Blue vs. Kasparov of our time," AI (artificial intelligence) researcher Randy Olson said of AlphaGo's victory on Twitter today. "This is huge." Olson, a University of Pennsylvania postdoctoral researcher specializing in artificial intelligence, machine learning and data visualization told us via e-mail today, "[A]lthough I'm obviously very excited about this victory for AlphaGo and the DeepMind team, it's important to keep our expectations grounded in reality."
While AlphaGo succeeded at a more complex game than did Deep Blue, it's still physically limited, just as even IBM's advanced cognitive computer (and Jeopardy winner) Watson is, Olson said.
"AlphaGo was literally born and bred to be the best Go player in the world, and that's it," he said. "Achievements such as this landmark victory for AlphaGo today are important to demonstrate how far the AI community has come, but for the most part we're still making expert machines that are very good at one task, but incompetent at all others."
That indicates that we're still not close to the point where AI reaches a tipping point that overtakes human intelligence, Olsen said. "[A]s an AI researcher myself, I'm still not convinced that we're on the cusp of a 'singularity' or any major technological breakthroughs in AI, but there are still many more interesting and useful applications of AI that we will see in the coming years," he said.
Image Credit: Google's AlphaGo challenge, against professional Go player Lee Sedol, via YouTube.
Posted: 2016-03-12 @ 10:36am PT
Update: AlphaGo today beat Lee Sedol in their third match, which means Google's AI bests the human champion 3-0. Details here: http://googleasiapacific.blogspot.co.uk/2016/03/alphagos-ultimate-challenge.html