To defeat Google’s AI, the humans must GO!

World Today

Go player Ke Jie of China, third from right, and other participants place pieces on a checkered cube during the opening ceremony of the Future of Go Summit in Wuzhen in eastern China’s Zhejiang Province, Tuesday, May 23, 2017. (AP Photo/Peng Peng)

For a second time Thursday, Google’s artificial intelligence computer has defeated China’s top player of GO, one of the last games machines have yet to master.

CGTN’s Joshua Barlow reports on the lead up to this unique match – and what the outcome could symbolize in the relationship between humans and machines.

To defeat Google’s AI, the humans must GO!

To defeat Google’s AI, the humans must GO!

At the Future of Go Summit in Wuzhen, China, a match is taking place between two formidable opponents: 19-year old prodigy and world champion Ke Jie - and Google's artificial intelligence system called AlphaGo.

At the Future of Go Summit in Wuzhen, China, a match is taking place between two formidable opponents:

KE JIE – a 19-year old Chinese GO prodigy who turned professional at age 11.
He reached the highest title of nine-dan rank in 2015 and is currently the world champion.

At the summit this week, Ke is playing against…

ALPHAGO – a computer program developed by Google’s DeepMind. AlphaGo is powered by some of the world’s most advanced technology, and uses artificial intelligence (A.I.) to learn gaming strategy and possible outcomes.

Known as Weiqi in China, GO was created during the Zhou dynasty more than 25 centuries ago. In the game, players take turns placing white or black stones on a 19*19 grid. The goal is to capture territory and opponent pieces.

It’s considered one of the world’s most complex games because of the near infinite combinations for position and strategy. According to Google, there are more potential positions in GO than atoms in the visible universe.

Computer experts believe by teaching AI to play games competitively – and learn from experience – it will develop capabilities to help humans solve larger problems.

Explaining the science behind AlphaGo

Jana Eggers, the CEO of Nara Logics, an artificial intelligence consulting company, explains how a computer defeated a human leading Go player.

A.I. is already used for everything from search engines and smart phones – to DNA mapping and weather prediction.

Still, winning competitively at strategy games is considered a true benchmark for A.I. – and there have been a handful over the last two decades.

In 1997, IBM’s Deep Blue system was the first A.I. to defeat a world chess champion, Garry Kasparov. And, in 2011, IBM’s Watson defeated human champions Ken Jennings and Brad Rutter in a highly publicized game on the TV quiz show Jeopardy!.

IBM Watson on Jeopardy!

IBM Watson wins big on Jeopardy! in 2011.

But, because of GO’s complexity, many believed it would be decades before AI could defeat a human.

That was until last year, when AlphaGo defeated Korean grand master and then world champion Lee Sedo in a 4-1 match. AlphaGo also defeated Ke Jie, in an unofficial online game in January.

Before this month’s match in Wuzhen, Ke said he was confident to beat AlphaGo.

But, after a first game lasting four and a half hours, AlphaGo beat Ke by a half-point – the closest margin for victory.

“The first time [we played] I thought it played a lot like a human,” said Ke. “Now I feel like his game is more and more like the ‘Go god’. Really, it is brilliant.”

Ke Jie’s hope for a comeback was dashed on Thursday, despite playing what Google’s AlphaGo indicated was the best game any opponent has played against it, said Demis Hassabis, founder of the company that developed the program.

AlphaGo “thought that Ke Jie played perfectly” for the first 50 moves, Hassabis said at a news conference.

“For the first roughly 100 moves, it is the closest game we have ever seen anyone play against the master version of AlphaGo,” he said.

Ke said the computer made unexpected moves after playing more methodically on Tuesday.

“From the perspective of human beings, it stretched a little bit and I was surprised at some points,” he said.

AlphaGo will compete with Ke Jie one more time this week, as well as take part in team play with human partners.

Many predict AlphaGo will be triumphant in the remaining exhibition games.

Ke Jie

Ke Jie competes with Google’s AlphaGo during the first GO match game in Wuzhen, China on May 23, 2017. (PHOTO: Google)

Some, like British physicist Stephen Hawking, warn AI may eventually destroy the human race. But others see A.I. as just another tool of evolution.

Zheng Hong, a retired ninth-dan master, has a more optimistic view of the game’s relationship to machine learning.

“It will not be the end of the world if AlphaGo wins,” Zheng said. “No man can outrun a car, but it still means a lot to get an Olympic medal as the fastest running human.”

Ke Jie, however, is determined to prevail.

“AlphaGo is a cold machine, and I cannot see its passion,” Ke said. “I will use all my passion to have the final fight against it.”

Ke Jie and AlphaGo will play their third and final game Saturday.

Story includes content from the Associated Press.