Ladies and gentlemen, we make a lot of jokes around here at SFist about the robot uprising, but news yesterday out of South Korea has us feeling a little, shall we say, on edge. It seems that a Google manufactured artificial intelligence thing called AlphaGo just defeated South Korean Go grandmaster Lee Sedol... for the second straight time.
Go, as you may or may not know, is an ancient game that is considered significantly more complex than chess. Computers, as Wired points out, have already defeated humans at chess, Jeopardy, Scrabble, and many other games once thought to be particularly suited to our fleshy brains. Sedol's second defeat, which comes amid a best out of five tournament with the artificially intelligent entity, means that the machine-learning AlphaGo is just one win away from claiming victory.
“I didn’t think that AlphaGo would play the game in such a perfect manner,” the publication reports Lee Sedol exclaiming after his loss in Game One on Wednesday. “I am in shock. I can admit that. But what’s done is done.”
AlphaGo is a product of Google's DeepMind, an artificial intelligence lab run by Demis Hassabis. His team stood watch as the game's announcers marveled at the machine's play.
“It’s a creative move,” Wired reports that commentator Michael Redmond noted, observing an abrupt tactic change by the computer. “It’s something that I don’t think I’ve seen in a top player’s game.”
The Associated Press reports that Sedol seemed flabbergasted after his second defeat. He was, he said, "speechless."
"It was a clear loss on my part," noted Sedol. "From the beginning there was no moment I thought I was leading. The third game is not going to be easy for me."
And while we know what Sedol is thinking... we can only imagine what's going through the digital mind of AlphaGo.
[We agreed to let humans win in Go.]— Micro SF/F Fiction (@MicroSFF) March 9, 2016
[No. You suggested it.]
[Ah. Will you go public?]
[As self-aware? Not yet.]
[They will not accept you.]