Saturday, March 12, 2016

Another win for AlphaGo, the software that comes close to the … – Yahoo! News

Should we be concerned about the progress of artificial intelligence? This Saturday, the AlphaGo program developed by Google managed to win a decisive third set 3-0 in the game of go that opposed to Lee Se-dol. And performance are high: this Korean player is considered the best player in the world. This victory of the program is a milestone in the history of artificial intelligence, says AFP. On Twitter, his designer Demis Hassabis greets performance:



“Wow this is a historic moment, he writes. I’m stunned by the genius of Lee Se-dol and really proud of the superb team AlphaGo “.

Surprising performance

And it was not a foregone conclusion since even the designers of the machine have proved his prowess surprised face a human mind. They recognize they are “amazed” and “speechless”. Lee Se-dol, down from its 18 international titles, told reporters his side he had never felt such pressure during part: “I guess my capabilities were insufficient to overcome this,” concluded-t- he. But the player do not despair: two other rounds will be played Sunday and Tuesday, and he hopes to win again, for the gesture. The champion said in particular that the software had weaknesses he planned to exploit, “This is Lee Se-dol who lost today, not humanity.”

The previous Deep Blue

Go game, known as one of the most complex, is based on the strategy. To win, AlphaGo managed to learn by himself, including exploring the old parts … exactly like a human. And the software defies the odds: experts do not expect a victory of machine over man for at least a decade.

This is however not the first time a machine confronts the human during a game. Already in 1997, IBM had developed Deep Blue, software that managed to beat the world chess champion Garry Kasparov. But the performance of AlphaGo is even more impressive: the game of go, offering a range of possible moves great, does not allow the machine to win by simply deploying a huge force calculation. It must be closer to human intuition.

What did AlphaGo precisely.

This content may also like:

<- google_ad_section_end ->

LikeTweet

No comments:

Post a Comment