In case you missed it, Korean Go master Lee Sedol is locked in battle for the future of humanity with AlphaGo, Google’s board-game supercomputer — and as of match two, things aren’t looking so good. AlphaGo is up two matches to none, although Lee Se-dol hasn’t gone down easily in either.
AlphaGo is the first artificial intelligence program to beat a professional human player at Go, an ancient Chinese board game that is so mathematically complex researchers thought humans still had the edge.
Well, as of Wednesday, they don’t. AlphaGo narrowly beat Lee Sedol in their first match out of a five match battle. Tonight’s match was also close — AlphaGo is definitely giving Sedol, a 9-dan pro, a definite challenge each game. Se-dol and AlphaGo stayed neck and neck for most of the four hour match, with AlphaGo showing signs of genuine creativity in its play, making moves that were outside of simple, easily-programmed patterns. The match stayed close toward the end, but Sedol ran into time trouble in the end game — each player has a cumulative two hours to make their moves, and AlphaGo still had over 15 minutes on its clock when Se-dol was into his emergency time (forcing him to make each move in under 60 seconds). Still, he hung on all the way to the wire, although AlphaGo took a slight advantage with Sedol under time pressure and held it with the relentless of a cold, impersonal machine. Sedol frowned, sweated slightly, and eventually lost.
This is both good and bad news for humanity — we’re building incredible computers, but we’re also losing to the robots, something tech visionaries are afraid of (although Space X’s Elon Musk praised AlphaGo’s achievement on Wednesday.)
Match three starts at 11 p.m. Eastern on Thursday, and matches four and five will be at the same time on Saturday and Sunday. You can livestream each match on DeepMind’s YouTube channel.