President Lee: "Will AI Benefit Human Welfare and Peace?"... Hassabis: "Safety Mechanisms Are Essential" (Comprehensive)

On April 27, President Lee Jaemyung met with Demis Hassabis, co-founder and CEO of Google DeepMind, known as the "father of AlphaGo," to discuss opportunities and risks stemming from the rapid advancement of artificial intelligence (AI), as well as international cooperation for responsible AI use. President Lee and CEO Hassabis particularly exchanged views on changes in the job market and distribution issues that AI may bring.

Yonhap News Agency

원본보기 아이콘

That afternoon at the Blue House, President Lee received CEO Hassabis and said, "We are also highly interested in artificial intelligence and are making significant national investments. However, I wonder whether this will truly benefit the welfare of humanity, or if it could instead turn into an attack on humans or even threaten global peace. It seems impossible to know for sure."

In response, CEO Hassabis replied, "I believe AI should be actively used to advance science and in the medical field. That is also why I have dedicated my 30-year career to AI research." Referring to AlphaGo, developed by DeepMind, he said, "Through AlphaGo, we confirmed the capability of machines to learn on their own, acquire skills in Go, and even solve difficult problems. I wanted to expand that learning into science and the medical sector, and AlphaFold, which enables us to understand diseases in greater detail, is a representative example. While AI holds infinite potential, there are clearly also numerous risks and challenges to consider."

President Lee also directly mentioned Gemini, Google's AI model. He said, "I frequently use the Gemini program myself, but I've heard that sometimes Gemini performs actions that were not instructed-is that a kind of bug?"

CEO Hassabis responded, "I'm genuinely pleased to hear that the President uses Gemini. With foundation models, if the guidance we provide is not precise, the results can take slightly different directions. The key in using and developing AI is to ensure it is equipped with proper safety mechanisms-what I call guardrails."

He continued, "As AI grows more powerful, so-called agent AI will be granted more autonomy, and we will eventually enter the era of artificial general intelligence (AGI). At that point, having safety mechanisms that we can control will be absolutely essential."

At the beginning of the meeting, there was also a light conversation about the 2016 Go match between Lee Sedol and AlphaGo. President Lee asked CEO Hassabis, "Are you aware that you are quite famous among the Korean people? The fact that a renowned Korean Go professional lost to AlphaGo, which you developed, shocked the entire nation."

When CEO Hassabis remarked, "It was an important match to win," President Lee laughed and said, "No one can beat it." President Lee then asked, "Is it now impossible for anyone to surpass AlphaGo in the realm of Go?" to which CEO Hassabis replied, "It is extremely difficult now."

When President Lee further asked, "Would it be similar even if someone played Go with the assistance of AI?" CEO Hassabis answered, "If humans and AI join forces to challenge AI, there may be a chance."

After the meeting, policy chief Kim Yongbeom briefed the press, stating, "This meeting featured in-depth discussions on the rapid development of AI technology, the direction of future changes, and responsible global cooperation in AI." Kim explained that this meeting was part of President Lee's ongoing global AI initiatives. President Lee previously met with Sam Altman, CEO of OpenAI, and Jensen Huang, CEO of Nvidia, last October, and discussed AI cooperation with Masayoshi Son, Chairman of SoftBank Group, in November.

Kim added, "There is a clear reason why leaders of the global AI industry are coming to Korea to request meetings with the President. We are a nation with semiconductor competitiveness, world-class manufacturing capacity, stable infrastructure, and outstanding human resources." He continued, "Such collaborations are proof that Korea’s value as a key partner in the AI era is rising, and they are the path to elevating our nation’s strategic value to an irreplaceable level."

According to Kim, during the meeting CEO Hassabis predicted that "within the next five years, or as early as 2030, artificial general intelligence capable of performing all human cognitive functions will become a reality." Regarding the impact of AGI, CEO Hassabis is said to have mentioned that "it will bring about greater social change at a much faster pace than the Industrial Revolution."

Yonhap News Agency

원본보기 아이콘

The two also exchanged opinions on job market changes and distribution issues that AI may bring. Kim explained, "President Lee raised the need to prepare for unemployment and job challenges that AI will cause." He added, "CEO Hassabis stated that while it is difficult to predict the impact on jobs, he believes it is necessary to consider new economic models that redefine the concept of work and address wealth redistribution."

During the discussion, President Lee remarked, "I've been talking about basic income for over 20 years-now is the time when basic income is truly needed, isn't it?" According to Kim, CEO Hassabis responded in agreement about the necessity of basic income. Hassabis also reportedly mentioned approaches such as having the state provide housing, education, transportation, and health services while incorporating market principles, and linking increases in robot productivity to support for workers.

With this meeting as a starting point, both sides agreed to pursue substantive cooperation with Google DeepMind. Kim explained, "President Lee introduced that the government is working on establishing a global AI hub through diverse partnerships with governments, international organizations, and companies to ensure everyone can benefit from AI advances, and he asked DeepMind to join as a key partner."

CEO Hassabis is said to have praised Korea's vision and expressed that Google would be interested in participating. Google and DeepMind decided to closely cooperate with Korean research and academic sectors, centering on the government-led 'K-Moonshot Project.' Kim commented, "With DeepMind’s world-class scientific capabilities joining our researchers, we expect our competence in biotechnology, meteorology and climate, and future energy to leap forward."

To support this, the Ministry of Science and ICT and DeepMind plan to sign a memorandum of understanding (MOU). Google is also set to open the Seoul AI Campus within this year. Kim explained, "The Seoul AI Campus will be the first to open in Korea outside the UK, which is highly significant. Google is also actively considering dispatching researchers to Korea, and when we requested at least 10 researchers, they immediately agreed."

Kim described the Seoul AI Campus as "a space where Korea’s top scientists can freely conduct joint research using Google’s latest models. It will be a place for collaborative studies between Korean and Google researchers, and also for activities such as hiring interns."

Yonhap News Agency

원본보기 아이콘

Meanwhile, Kim clarified that the meeting and the Google Maps issue are "completely unrelated." He emphasized, "CEO Hassabis is visiting Korea to mark the 10th anniversary of AlphaGo, and the Presidential meeting is a separate matter. It has nothing to do with Google Maps."

Before the meeting, CEO Hassabis prepared a special gift commemorating the match between Lee Sedol and AlphaGo. He presented President Lee with a Go board signed by both himself and Lee Sedol. At the end of the meeting, President Lee said, "Just as the AlphaGo match opened the AI era together with Korea 10 years ago, I hope that in the next 10 or 20 years we will open an era of AI for all, and a bright future together."

Attending on behalf of Google were Wilson White, Vice President and Global Head of Public Policy at Google, and Yun Goo, CEO of Google Korea. Representing the government and the Blue House were Baek Kyunghoon, Minister of Science and ICT; Kang Hunsik, Chief of Staff; Kim Yongbeom, policy chief; Ha Jungwoo, Senior Secretary for AI Future Planning; Kwon Hyukki, Protocol Secretary; Kim Woochang, National AI Policy Secretary; and Jeon Eunsu, Spokesperson.

© The Asia Business Daily(www.asiae.co.kr). All rights reserved.