[✏️ Full version for review] “I thought physical labor would be safe”: AI is changing jobs—where should we invest? | Director Hong Ki-bin

Director Hong Gi-bin explains that large language models have rapidly absorbed human linguistic and logical knowledge, but reaching AGI requires world models and physical AI that understand causality in the physical world. The key bottleneck is capturing as data the bodily movements and sensations that skilled workers find difficult to explain verbally—in other words, tacit knowledge. Countries are collecting work data through sensors and video, but workers have little incentive to cooperate because they risk losing their jobs after handing over skills accumulated throughout their lives. He therefore argues that rather than purchasing movement data as a one-off transaction, governments, companies, and labor unions should establish a new social compact to manage it as a shared asset and ensure that workers receive ongoing income and rights when robots use it. He acknowledges that China and other countries could make the transition quickly while South Korea spends time reaching agreement, but believes that sufficient agreement can reduce implementation costs and conflict and may prove more efficient in the long run. The AI transition requires long-term national capital accounts and social investment extending beyond short accounting cycles. He proposes that building a data commons from South Korea's broadcast video, cultural, administrative, land, and other data could provide a foundation for sovereign AI and new industrial infrastructure. From an investment perspective, robot hardware should be considered alongside high-quality movement and cultural data, long-term infrastructure, and frameworks for agreements and revenue sharing. Finally, he emphasizes that competition over AGI and ASI could absorb resources and capital at an exponential rate, so society must deliberate politically on the goals and limits of the technology.