Home > Opinion > Columns

AI governance: The missing link in national AI strategies

Published: 09 Jul. 2025, 00:01

Yeom Jae-ho

The author is the president of Taejae University.

In November 1970, the Oakland Tribune reported the failure of a U.S. Economic Development Administration project aimed at revitalizing the Oakland port area. The project sought to transform a former naval base into a commercial hub to create jobs, but ultimately fell short. In their book "Implementation" (1984), political scientists Aaron Wildavsky and Jeffrey Pressman analyzed the reasons behind the failure. While stakeholders supported the plan during its formulation, they were passive during its execution. The takeaway: Policies must be designed with implementation in mind from the outset.

Today, the rise of AI presents a similarly complex policy challenge. Much like Christopher Columbus’s “discovery” of the New World, AI is reshaping the trajectory of human civilization. According to Google futurist Ray Kurzweil, AI will surpass human intelligence by 2029 and reach a point of “superintelligence” by 2045, pushing civilization toward a singularity.

President Lee Jae Myung takes part in a commemorative ceremony at the launch event for the Ulsan AI Data Center, titled “Korea’s AI Highway,” held at the Ulsan Exhibition and Convention Center on June 20. From left: Industry Minister Ahn Duk-geun, National Assembly Science, ICT, Broadcasting and Communications Committee Chair Choi Min-hee, SK Group Chairman Chey Tae-won, President Lee, Prasad Kalyanaraman, Vice President of Infrastructure at Amazon Web Services, Science and ICT Minister Yoo Sang-im and Ulsan Mayor Kim Doo-gyeom. [JOINT PRESS CORPS]

National strategies for AI typically focus on three pillars: computational infrastructure, talent and data. The United States is building a megascale computing center through the Stargate Project, with OpenAI, Oracle and SoftBank committing as much as 730 trillion won ($534 billion). The European Union is also investing 300 trillion won in AI infrastructure development plan. France has pledged 163 trillion won to its AI data centers. Singapore has gone as far as to offer approximately 6.7 million won in monthly stipends to Ph.D. students in AI programs, regardless of nationality. China, relatively unburdened by privacy constraints, has developed its high-performing DeepSeek model by leveraging massive datasets.

Korea, too, is stepping in. The government recently announced a plan to invest 100 trillion won to build a sovereign AI computing center.

Then-presidential contender Lee Jae Myung of the Democratic Party holds up a FuriosaAI NPU chip during a visit to the company’s office in Gangnam District, Seoul, on April 14. [NEWS1]

During a recent visit to Shanghai and Hangzhou as part of the Korean Peninsula Peace Odyssey, I met with AI firms and researchers at Zhejiang University, the birthplace of DeepSeek. One company emphasized a critical point: Even more important than computing power, data and talent is AI governance. In other words, social systems — more than technical capability — will determine AI’s real-world impact.

Much of today’s social structure stems from the mass production systems of the 20th century. Back then, large organizations thrived on standardized, repetitive work. Bureaucracy expanded in both manufacturing and office-based functions like human resources and finance. But AI is poised to replace much of this routine labor, while robots take over physical processes. This transformation will reach across sectors — from law and medicine to education, finance and the arts.

A nation’s AI competitiveness, therefore, will hinge less on raw technology and more on practical integration. For AI to enhance productivity and shift human labor toward more creative tasks, existing data and workflows must be opened up for machine learning. Resistance to this transition — especially the withholding of essential data — will render even the most advanced AI tools ineffective.

Take health care. Korea has comprehensive national medical data, yet if hospitals refuse to share it due to privacy concerns, medical AI development stalls. Legal precedents are digitally archived, but restricted access limits their usefulness for AI-based legal analysis. In manufacturing, fear of information leakage may prevent companies from providing data necessary for AI-driven productivity gains.

The building housing the headquarters of Chinese AI startup DeepSeek is seen in Hangzhou in China's eastern Zhejiang province on January 28. Fears of upheaval in the AI gold rush rocked Wall Street following the emergence of a popular ChatGPT-like model from China, with U.S. President Donald Trump saying it was a ″wake-up call″ for Silicon Valley. [AFP/YONHAP]

This is why governance must take precedence. Without a regulatory framework that enables AI to learn from existing data, investments in infrastructure and sovereign language models risk being wasted. The United States and China are advancing in AI not solely because of resources, but because they’ve built governance models that facilitate data access and usage.

The Korean government, which has emphasized pragmatism and AI-driven national competitiveness, must prioritize the establishment of effective AI governance. Without policies that dismantle vested interests and enable AI to be deployed at scale, massive investments in technology may yield little public benefit.

Translated from the JoongAng Ilbo using generative AI and edited by Korea JoongAng Daily staff.