When IBM's Deep Blue beat chess champion Garry Kasparov back in 1997, the world was agog over AI's potential. This week, Google DeepMind's AlphaGo system will face an even tougher test in a series of matches against a top-ranked master in the ancient game of Go.
The tournament will in many ways be an ultimate challenge for the artificial-intelligence system following its victory last fall -- roughly a decade earlier than anyone had expected -- over European champion Fan Hui.
Go is an ancient Chinese game widely considered even more complex than chess, even though its rules are simpler. The player’s objective in the game is to surround the opponent’s pieces by alternately placing black and white pieces on a 19-by-19-line grid while simultaneously avoiding having his or her own pieces surrounded. With more possible positions than there are atoms in the universe, Go has long been considered an ultimate challenge for AI researchers.
AlphaGo taps neural networks and advanced “tree search” programs. It also takes a different approach to the game than had been tried previously. It's done remarkably well. The system has won more than 99 percent of the games it’s played against the strongest other Go programs. In October, it beat Fan Hui 5-0, making it the first computer program ever to beat a human professional player in a nonhandicapped game.
AlphaGo "plays like a human," Fan Hui said in an interview early this year.
Now, in a series of five matches starting this week, AlphaGo will take on Lee Se-dol, the 33-year-old South Korean professional who is among the world's top Go players. The games will be even, with no handicaps; the winner will get $1 million in prize money. If AlphaGo wins, the proceeds will be donated to UNICEF, STEM and Go charities, Google DeepMind said.
The matches will be held in Seoul starting at 1 p.m. local time on Wednesday, Thursday, Saturday, Sunday and next Tuesday, which translates to 11 p.m. on the previous day in U.S. Eastern Time. They will be streamed live on DeepMind’s YouTube channel as well as broadcast on TV throughout Asia. Each match is expected to last between four and five hours.
"I heard Google DeepMind's AI is surprisingly strong and getting stronger, but I am confident that I can win, at least this time," Se-dol said.
Which player wins could have implications well beyond the world of Go. With Go widely considered a final frontier for dominance by the human intellect, an AI victory will surely spark more than a few dire headlines about silicon's inevitable rise. That's a topic that generates all kinds of angst about job security, in particular.
Many experts, however, aren't so concerned. AI may be well-equipped to win at Go, but as Shashi Upadhyay, CEO of AI startup Lattice Engines, recently said, “Humanity has a lot of things to be proud of — being ‘Go’ champion doesn’t need to be one of them.”
And, in the words of H2O.ai cofounder and CEO SriSatish Ambati, AI may well end up freeing us "to be humans again."
Stay tuned for more coverage as the tournament unfolds.