Computers just got even more scarily smart. A program designed by Google (GOOG[1]) researchers has become the first to defeat a professional human player at the ancient Asian game of Go.
Google Deepmind's program, AlphaGo[2], recently crushed the current European Go champion, Fan Hui, by five games to nothing. Alp haGo's victory -- a feat that experts had believed would take at least another decade to achieve -- is detailed in an article[3] in the science journal Nature this week.
Enabling computers to master Go has been a kind of holy grail[4] for some artificial intelligence scientists.
Back in the '90s, software programs became adept at classic board games l ike backgammon. Their rapid progress culminated in the historic victory of IBM's Deep Blue computer over world chess champion Gary Kasparov in 1997.
It's taken almost another 20 years for artificial intelligence to get to grips with the mind-boggling complexities of Go, which experts say relies deeply on intuition. Until recently, software programs could only compete with human amateurs.
Related: Facebook CEO says nothing to fear from artificial intelligence[5]
"The number of possible configurations of the board is more than the number of atoms in the universe," said Demis Hassabis, a researcher at Google DeepMind[6] in London, where AlphaGo was developed. One of the program's key traits is the ability to learn from games against itself.
Believed to have originated in ancient China, Go has been played by people across East Asia for thousands of years.
The rules are simple: two opponents try to take territorial control of a board by taking turns at placing their pieces on a grid of lines. They can capture each others' pieces by surrounding them.
But the range of potential sce narios is vast.
"In chess, the number of possible moves is about 20 for the average position," Hassabis said. "In Go, it's about 200."
Related: Tech heavyweights invest $1 billion in artificial intelligence[7]
Next up for the skillful program is a showdown in March in South Korea with Lee Sedol, the dominant player of the past decade whom Hassabis describes as "the Roger Federer of Go."
The researchers say they expect AlphaGo's attributes to eventually be put to use beyond the realm of board games in areas like Google's own apps, and even medicine.
References
- ^ GOOG (money.cnn.com)
- ^ AlphaGo (deepmind.com)
- ^ an article (www.nature.com)
^a kind of holy grail (www.wired.com)- ^ Related: Facebook CEO says nothing to fear from artificial intelligence (money.cnn.com)
- ^ Google DeepMind (deepmind.com)
- ^ Related: Tech heavyweights invest $1 billion in artificial intelligence (money.cnn.com)
