Page 23 - ATODAY

TECHNOLOGY A23
Wednesday 9 March 2016

Go master: AI will one day prevail but beauty of Go remains

YOUKYUNG LEE South Korean professional Go player Lee Sedol, center, poses for the media with Eric Schmidt, executive chairman of Alphabet,
AP Technology Writer right, and ceo of Google DeepMind Demis Hassabis, left, after a press conference ahead of the Google DeepMind Challenge
SEOUL, South Korea (AP) — Match in Seoul, South Korea, Tuesday, March 8, 2016.
Computers eventually will Lee Sedol and Google’s artificial intelligence program, AlphaGo will play in a five game matches from March 9 to March 15.
defeat human players of
Go, but the beauty of the Associated Press
ancient Chinese game of
strategy that has fascinat- man game, it is important sional Go player. long-term results of each the game is not all about
ed people for thousands to read the other person’s That changed last year move and predict the win- victory.
of years will remain, the Go energy and force. But in when AlphaGo defeated a ner. Known as baduk in Korean
world champion said Tues- this match, it is impossible European Go champion in Using this approach, Al- and weiqi in Chinese, Go is
day. to read such things. It could a closed-door match later phaGo beat the European more than a game in Asia.
South Korean Lee Sedol, a feel like I’m playing alone,” published in the journal Na- Go champion by searching Players’ moves reflect their
Go master who has won Lee said. ture. through far fewer positions personalities and distinc-
18 international titles since Because the number of Google’s DeepMind team than those a traditional AI tive styles, and the life-and-
he became a professional possible Go board positions created a system to narrow machine like DeepBlue, the death battles between
player at age 12, said the exceeds the number of down a vast search space famed IBM computer that black and white stones for
risk of human error means atoms in the universe, top of near-infinite possible se- defeated the world’s chess territory on the 19 by 19
he may not win his match players rely heavily on their quences of moves in the champion in 1997, would square grid are often used
this week against Google’s intuition, said Demis Has- game. have to consider, Hassabis to illustrate important life
artificial intelligence ma- sabis who heads Google’s AlphaGo was first trained said. lessons.
chine, AlphaGo. DeepMind, the developer to mimic experts’ Go AlphaGo also has other “Of course I can lose. But a
“Because humans are hu- of AlphaGo. moves based on data from strengths as a machine. computer does not play by
man, they make mistakes,” This has made Go one of about 100,000 Go games “I think the advantage of understanding the beauty
the 33 year-old said a day the most complex games available online. Then it AlphaGo is that it will never of Go, the beauty of hu-
before the first of the five ever devised and the ulti- was programmed to play get tired and it will not get mans,” he said. “My job is
games he is due to play mate challenge for the AI against itself and “learn” intimidated either,” Hass- to play Go more beauti-
against AlphaGo. “If there experts, who had expect- from its mistakes. abis said. fully.”
are human mistakes, I ed that it would take at Lee said he hopes to hold That beauty, many Go fans
could lose.” least another decade for a The team also designed onto his title, but also wants believe, is something a ma-
It was Lee’s first admission computer to beat a profes- a system that enabled Al- to remind audiences that chine cannot replicate.
of his weakness against phaGo to anticipate the
Google’s AI machine and
also a dialing down of
his confidence from two
weeks ago, when he had
predicted a 5-0 result in his
favor.
After watching Google’s
presentation of how Al-
phaGo works, Lee said he
thought a machine might
be able to imitate human
intuition,
even though the intuition
may not be as sharp as a
person’s.
A loss for Lee would be a
historic moment for the AI
community.
Human errors are not his
only vulnerability.
Lee said that in playing
against a machine, the
absence of visual cues
that human players use
to read the reactions and
psychology of their oppo-
nents puts him in unfamiliar
territory.
“In a human versus hu-