Apple News Facebook Twitter 新浪微博 Instagram YouTube Wednesday, Mar 15, 2023
Search
Archive
English>>

AlphaGo shows its total board game dominance by thrashing

(Global Times)    09:09, January 11, 2017

Cho Chin-kun, a 60-year-old South Korean professional Go master, plays a game with Japanese artificial intelligence program DeepZenGo in Tokyo on November 23, 2016. Cho won the three-game match at 2-1. Photo: IC

  Cho Chin-kun, a 60-year-old South Korean professional Go master, plays a game with Japanese artificial intelligence program DeepZenGo in Tokyo on November 23, 2016. Cho won the three-game match at 2-1. Photo: IC

AlphaGo, the AI player of the Go board game, hit the headlines again recently by achieving 60 straight online victories over more than 5o of Asia's top Go players.

AlphaGo, under the username "Master," has defeated 59 world-class Go players on online Go platforms since December 29, 2016, including players from China, Japan and South Korea, such as Ke Jie, Yuta Iyama and Park Jeong-hwan. "Master" revealed that it is an AI on January 4, before a match with world-class Chinese player Gu Li in which it easily achieved its 60th victory.

"Thanks for the shock AlphaGo brought to the game of Go ... though I knew the identity of Master from the beginning, I still strongly wished that a human could win when watching the games," Ke Jie wrote on his Sina Weibo account on January 4.

Man behind the machine

Aja Huang, senior research scientist at Google DeepMind and the lead programmer of AlphaGo, was the one who moved the game pieces for AlphaGo during the matches.

His computer science and information engineering master's tutor at National Taiwan Normal University, Lin Shun-Shii, told the Global Times about Huang and AlphaGo on Thursday.

Huang started to learn Go when he was a child and continued playing into adulthood, which helped him develop AlphaGo, said Lin.

In around 2004, Huang began to develop a Go AI named after his wife Erica. Huang observed other similar programs and often discussed how to improve the Erica AI with Lin.

At the beginning, at competitions which pit Go programs against each other, Erica was an average player, but after gradual improvements, Erica became one of the best programs at the time and won Huang a gold medal at a Japanese competition.

Today, AlphaGo is still using some of the technologies pioneered in Erica.

Lin told the Global Times that AlphaGo was upgraded after playing historic matches against some top Go players in March 2016, when it was "not so good sometimes."

However, the AI won all 60 matches this time, which "never happened in history."

"No other program but AlphaGo could win 60 matches, so DeepMind, the developer, admitted that 'Master' was AlphaGo," Lin noted.

"Now that our unofficial testing is complete, we're looking forward to playing some official, full-length games later this year in collaboration with Go organizations and experts, to explore the profound mysteries of the game further in this spirit of mutual enlightenment," DeepMind announced on January 4.

Humans besting humans

Nowadays, with the rapid development of AI, people have mixed feelings about the technology - both hopes and fears.

Lin believed that "it is an inevitable trend that humans cannot beat computers anymore."

Our computer systems have become smarter and more powerful, and if the computers are better than you at the most advanced decisions, just let them do it, Lin said. "You cannot help it."

"This is a victory for human beings, not for machines," said Yu Kai, founder of the Institute of Deep Learning at Baidu Inc and CEO of Horizon Robotics, the Xinhua News Agency reported.

Yu also argues that as the algorithms controlling machines are created by humans, so if their orders are changed, the machines cannot win.

Yu believed that AI could become partners to humans and be used for production, education, medicine and transportation. The development of AI technology does not aim to replace mankind's with machines, but to extend human's abilities, he says. "Technologies could help humans in the areas that we are not good at."

When an AI surpasses humans in some areas, it is actually human beings surpassing themselves, said Wang Yuefei, chief of the Beijing-based State Key Laboratory of Management and Control for Complex Systems, according to Xinhua.

"How could we benefit more people with these advances is the question that we should consider the most," Wang noted.

DeepMind announced on January 3 that it would expand its business into the fields of medical treatment and video games, people.cn reported on Saturday.

(For the latest China news, Please follow People's Daily on Twitter and Facebook)(Web editor: Kou Jie, Bianji)

Add your comment

We Recommend

Most Read

Key Words