Sat, Nov 11, 2017 - Page 3 News List

AI not a threat to people: forum

By Lin Chia-nan  /  Staff reporter

Artificial Intelligence (AI) can serve as an instrument of humankind, rather than a threat, Aja Huang (黃士傑) — one of the key developers of Google’s AI program AlphaGo — said yesterday at the first Taiwan AI conference in Taipei.

Huang talked about the most impressive moments in the program’s evolution at the two-day conference hosted by Taiwan Data Science Foundation and Academia Sinica’s Institute of Information Science.

AlphaGo, a computer program developed by Google’s DeepMind Technologies, has revolutionized ideas about machine learning after it beat Lee Sedol, a South Korean professional 9 dan ranked player, 4-1 in March last year.

It claimed an even more significant victory when it defeated the world’s best go player since 2014, Ke Jie (柯潔), in May.

Its triumph over Lee is a small step for the team, but a giant step for humankind, Huang said yesterday, quoting Demis Hassabis, the founder and chief executive officer of DeepMind Technologies.

Recalling the game with Ke in Wuzhen, China, Huang said it turned out to be a pleasant and constructive experience.

“The two games in South Korea and in China are different, in that Ke Jie was smiling during the game,” Huang said. “In contrast, [the game with] Lee was extremely stressful, because he seemed to be fighting [the AI program] on behalf of humanity.”

While the original AlphaGo was developed by learning from human experiences, its more advanced version, AlphaGo Zero, which was launched last month, is said to have achieved superhuman proficiency based solely on reinforcement learning without the input of human data.

Their research was published in a paper titled “Mastering the game of Go without human knowledge” in the journal Nature on Oct. 18.

AlphaGo Zero has outsmarted all of its past learning, Huang said, adding: “It is a trend, and i cannot stop its growth.”

Nonetheless, AI is not being developed to combat humankind, but can serve as an instrument, Huang added.

Also speaking at yesterday’s conference, Minister of Science and Technology Chen Liang-gee (陳良基) said the ministry is to allocate NT$5 billion in five years to establish four AI innovative research centers at National Taiwan University, National Tsing Hua University, National Chiao Tung University and National Cheng Kung University, with the aim of cultivating about 3,000 researchers each year.

In addition to computers, self-driving cars, robots and data science are key application areas for AI, while the ministry is to build two AI robot maker spaces in Taichung and Tainan, he said.

It will also propose a bill on unmanned vehicles to the Legislative Yuan by the end of this year, he added.

Trend Micro Inc research and development department senior architect Daniel Cheng said he welcomes the government’s assistance in AI development, even though the company had started five or six years ago.

The government should provide businesses with more open data, which is essential to the development of AI, he said.

In related news, Academia Sinica has established a Taiwan AI Academy, with Harvard computer science professor Kung Hsiang-tsung (孔祥重) as president.

Application for the school is open for university graduates until Dec. 4.

This story has been viewed 3377 times.

Comments will be moderated. Remarks containing abusive and obscene language, personal attacks of any kind or promotion will be removed and the user banned.

TOP top