“I don’t really know if it’s a good or bad move,” said Michael Redmond, a commentator on a live English broadcast. “It’s a very strange move.” Redmond, one of the Western world’s best Go players, could only crack a smile.
“I thought it was a mistake,” his broadcast partner, Chris Garlock, said with a laugh.
Sedol, however, was more serious. He stared at the board, then got up from the table and left the room.
As Sedol returned after a few minutes and pondered his next move, it became clear that AlphaGo’s move was no mistake. It might be strange, but it definitely wasn’t bad. It was brilliant.
Sedol would take almost 16 minutes to make his next move. He would never recover, losing the match.
“Almost no human pro would’ve thought of it, I think,” Redmond said after the match.
AlphaGo’s move in the board game, in which players place stones to collect territory, was so brilliant that lesser minds — in this case humans — couldn’t initially appreciate it. The move also opened a debate about whether increasingly powerful machines have mastered creativity, a trait widely thought to be strictly in the domain of humanity.
Pedro Domingos, a computer science professor at the University of Washington and author of “The Master Algorithm,” saw a parallel between AlphaGo’s style and how chess prodigy Bobby Fischer was feared because his early moves were considered too foolish to even be made. But as Fischer’s matches wore on, the ill-advised moves suddenly looked genius.
“If that’s not creative, then what is?” Domingos asked.
He sees machines delivering creative results, and they’re just getting started. Domingos believes a computer eventually will write a best-selling book. And he thinks there’s a 50-50 chance that a computer writes a hit pop song in the next decade, given advances in artificial intelligence techniques and computing power.
Domingos said such advances shouldn’t come as a surprise, as machines increasingly demonstrate that creativity isn’t magical and distinctly human.
“We seem to have this mythical view that ‘oh, it just pops into my head, and it’s magic,’ but it’s not,” Domingos said. “Human beings [like machines] are also creative because of this massive parallel search learning process that goes on in their brains.”
For others, AlphaGo’s dominance doesn’t deserve the label of creativity.
“This is the latest in a long history of overhyped [artificial intelligence] demonstrations,” said Jerry Kaplan, a computer scientist and author of “Humans Need Not Apply: A Guide to Wealth and Work in the Age of Artificial Intelligence.” “It’s a machine engaging in taking a certain set of actions that are a result of the clever programming that has been done to build it.”
Kaplan mentioned IBM Deep Blue’s victory over chess grandmaster Garry Kasparov in 1997, and IBM Watson’s win over Ken Jennings in Jeopardy as examples of artificial intelligence hype. While heralded with much media attention, far-ranging implications from those victories haven’t been felt in society.
DeepMind, the London-based artificial intelligence group that Google owns, made unexpected progress in Go because of its use of a technique called deep learning, which excels at recognizing patterns in visual images. Deep learning is in vogue now, and led to developments in other areas such as autonomous vehicles, and interpreting medical images.
“Right now they have a new hammer and everyone in Silicon Valley is swinging it at what looks vaguely like a nail,” Kaplan said.
While experts say AlphaGo’s decisive victory is indeed a milestone, arriving years ahead of expectations, they also put the win in perspective.
“AlphaGo does not advance the state of the art in machine learning,” Domingos said. “There is nothing fundamentally new about the kind of machine learning that is going on there. It’s more that they pulled off something amazing with it.”
Earlier in his career Domingos himself tried to build computers capable of beating humans at Go. He credited Google’s team with realizing it could apply the trendy tool of deep learning to the 2,500-year-old game.
After the series ended DeepMind chief executive Demis Hassabis said techniques like AlphaGo show much promise, but are still in their early days. He said the group wasn’t sure what it would do next with AlphaGo, but reiterated his interest in applying the tool to scientific discoveries and medicine. In February Hassabis announced DeepMind Health, a partnership with Britain’s National Health Service to build technologies to help care for patients.
He also encouraged artificial intelligence researchers to stress ethics, as some have warned of the potential catastrophic potential of artificial intelligence to one day cause mass unemployment or even human extinction.
“As with all powerful technologies they bring opportunities and challenges,” Hassabis said. “We have to make sure that developers of these kind of systems — all AI researchers around the world — think about the ethical responsibilities they have to build these systems in the right way and to deploy them for the right purposes.”