South Korea fears losing race for AI leadership, Naver's AI expert says

20 August 2024
By Choonsik Yoo

South Korea started earlier than others in the development of artificial intelligence, but its status is now threatened as competition among countries outside the US and China intensifies to develop AI ecosystems that best serve local needs, South Korea’s most influential AI expert at tech giant Naver warned.

Ha Jung-woo, who heads Naver's AI development operations, recommended that the government and parliament focus on legislating lower-level laws first to promote the AI industry and educate users, rather than spending time and energy trying to enact an all-encompassing basic AI law.

In an interview with MLex, Ha also said that South Korea is on the verge of losing the race to develop so-called sovereign AI ecosystems due to the limited scale of state and private investment, while other countries are implementing multi-billion-dollar spending plans.

“Awakened by ChatGPT, governments of several countries have been offering large-scale support measures for their local industries since 2023,” Ha said, referring to the explosive growth of the AI industry sparked by OpenAI’s release of ChatGPT 3 in November 2022.

He listed France, Germany, Japan and Canada as examples of major countries that are trying to catch up with early movers such as South Korea by offering direct and indirect support packages valued in the billions of dollars, whereas South Korea is hesitant or unable to offer direct investment.

Promotion before regulation

South Korea sees the AI boom as an indispensable opportunity to revive its economy, which faces severe threats from an aging population and signs of peaking productivity growth, and has made it clear that its policy would prioritize promoting innovation over regulating risks.

However, the situation has changed significantly due to the faster-than-expected spread of AI systems in everyday life around the world, especially after the European Union enacted the AI Act this year, which contains a series of prescriptive restrictions.

While acknowledging the EU AI Act's benchmark effect for other countries, Ha warned against modeling South Korea’s AI governance framework after it, emphasizing the “completely different” social, historical, political, legal and cultural situation facing the country.

Sovereign AI as business model

Naver, South Korea’s dominant one-stop provider of services that Google, Amazon and Facebook specialize in individually, has long been pushing for the concept of "sovereign AI" — a nation’s capability to produce AI using its own infrastructure, data, workforce and business networks.

“Artificial intelligence is basically a projection of certain values, and those values come from data, and are determined by where the data originates,” Ha said, adding that global AI models are mostly trained on data produced in the US and naturally project the values of users in the US.

Naver is one of the few companies outside the US and China that has succeeded in developing its own AI models running on its own cloud infrastructure. It is now working to sell these business models to third-world countries concerned about technological dependence on the US or China.

He noted that Naver’s capabilities across the entire process — ranging from cloud services to custom-tailored AI systems — have already helped it win contracts in countries such as Saudi Arabia that are worried about being too dependent on US or Chinese technologies.

National AI committee

Ha expressed high hopes for the country’s national AI committee, which is due to be established by early next month, but said more experts and industry stakeholders need to be given the opportunity to present their opinions to the committee.

He said that he had no chance to provide his opinion during the preparatory process for the committee's establishment, which will be chaired by the country’s president, while the head of a local university is set to be the sole deputy chairperson.

Along with the need for the government to help funnel massive investment into the AI industry, Ha also said the country was largely successful in setting up plans to train experts in AI technology earlier than many other countries, but is failing to retain that talent.

“We were very fast in implementing projects to train people across a wide range of subjects from as early as 2019,” Ha said. “The talent trained through such efforts should be contributing to making the country more competitive, but they are leaving the country.”

For the latest developments in AI & tech regulation, privacy and cybersecurity, online safety, content moderation and more, activate your instant trial of MLex today.