South Korea's AI Leadership Requires Human-Centered Research

When discussing artificial intelligence competitiveness, we often think first of larger models, more semiconductors, and faster commercialization. However, a nation's AI competitiveness cannot be achieved through cutting-edge industrial technology alone. A country becomes a true AI powerhouse only when research that solves difficult social problems and improves people's lives accompanies commercial development—even if such research does not generate immediate profits.

It is therefore particularly noteworthy that researchers from the Korea Advanced Institute of Science and Technology (KAIST) have recently received best paper awards at leading global conferences for their "AI for people" research.

One study, in which I participated, focused on slum detection and was selected as the best paper in the AI for Social Impact track at AAAI, a premier AI academic conference. The other, by the research team of Professors Oh Hye-yeon and Kim Ju-ho, won the best paper award at CHI, the most prestigious international conference in human-computer interaction.

The slum detection research, conducted jointly with Professor Cha Mi-young (KAIST and Max Planck Institute) and Professor Yang Jae-seok of Chonnam National University, presented technology using satellite imagery and AI to identify slum areas in developing-country cities. Slums are spaces where urban poverty is most intensely concentrated, yet many countries cannot properly determine where or how extensively they have formed. Our research team developed detection technology applicable to multiple cities by learning the characteristics of slums, which vary by region in shape, density patterns, and building materials.

The significance of this research lies not in the technology itself but in its ability to capture urban poverty and inequality that macro statistics fail to reveal, and to provide a basis for policy and resource allocation even in data-scarce regions. In other words, this is AI that shows where public intervention is most urgently needed, rather than improving advertising efficiency.

The CHI award-winning research by Professors Oh and Kim's team demonstrates the same concern. This study introduced a large language model-based writing support tool into middle school English classrooms for six weeks and analyzed how 157 students used it. The results showed that while AI's step-by-step hints benefited some students, they also increased dependency and frustration among lower-performing students. Additionally, as students produced similar-quality writing with AI assistance, teachers found it harder to identify which students actually needed help.

Rather than the optimistic view that "using AI automatically makes things better," this research directly examined in the classroom who benefits and who faces new barriers.

Though different in field and technology, both studies share the distinction of directly addressing human and social issues and gaining recognition for their value on the world stage. This demonstrates that Korean researchers are world-class not merely in keeping pace with industrial AI competition, but in answering the more fundamental questions of where, how, and why AI should be used.

A characteristic of such research is that it takes place over extended periods for the benefit of society as a whole. It requires data development and field collaboration, and results do not immediately translate into revenue. This is why national support and the roles of universities and public research institutions are crucial.

It was encouraging that the National AI Committee last month outlined directions for public utilization of AI and social transformation in its "AI Action Plan." However, "AI for people" does not grow through declarations alone—stable financial support must follow.

More importantly, such investment is not only beneficial for the public good but can also represent a strategic opportunity for Korea. We are a country that combines rapid digital transformation, high educational enthusiasm, a comprehensive public system, and experience in international development cooperation. These conditions provide fertile ground for testing and advancing inclusive, public-oriented AI.

At this moment when the world is questioning the direction of AI competition, Korea can become a country that demonstrates the meaning and direction of AI for people first, rather than one that merely follows technologies created by others.

Our researchers are already proving this possibility. What is needed now is support that enables more such efforts and national vision that properly recognizes their value. If we can produce world-leading results in AI for people, that will be Korea's truly sustainable AI competitiveness.