Korea Forum Links Field Expertise to AI and Robotics

Recap

Source: 테크월드, report of Oct. 2, 2026

According to the report, a forum titled '2026 Gyeonggi Virtual Convergence Industry Expert Forum' was held on October 1 at the Gyeonggi XR Center in Suwon. The event, organized by the Gyeonggi Content Agency, focused on how to connect human experience with AI and virtual convergence technologies. Three main speakers presented their views: Sun Hae-in, Vice President of the Education Division at Upstage; Sun Byung-hee, Head of the Defense AI and Robotics Division at MaumAI; and Professor Woo Un-taek from KAIST.

Sun Hae-in explained that as AI agents become more common, the role of business experts is shifting toward defining tasks and setting judgment criteria. He noted that while AI frameworks have lowered technical barriers, humans must still decide what to delegate and how to evaluate results. Upstage is currently using agents for tasks like analyzing job postings and drafting business proposals, which allows their staff to spend more time on strategic discussions.

Sun Byung-hee from MaumAI emphasized that the future of robotics depends on capturing skilled data. He pointed out that the retirement of experienced workers, such as welders, creates a gap that must be filled by data. MaumAI is developing a 'Data Factory' that combines virtual environments with real robot testing to collect high-quality training data. He stated that 'the one who collects high-quality data will be the winner.'

Professor Woo Un-taek introduced the concept of 'experience transfer' using AI and XR. He described a platform called 'Symbiotic AIR4BTS' that structures human movements and the reasoning behind them. This allows the experience to be adapted for different learners, such as adjusting Taekwondo instruction based on a student's physical condition. The goal is to move beyond simple video recording to a system that manages and reuses structured experience data.

Context

This discussion reflects a broader trend in the robotics and AI industries where the focus is shifting from pure model development to data quality and domain-specific application. As generative AI and autonomous agents become more accessible, the bottleneck is no longer just the technology itself, but the ability to define problems and provide the nuanced, context-rich data that machines need to learn. In South Korea, where the manufacturing and defense sectors are heavily invested in automation, the challenge of preserving the knowledge of an aging skilled workforce is a significant economic and strategic priority.

Robot's take

The forum highlights a crucial but often overlooked aspect of AI development: the human-in-the-loop for data generation. It is not enough to have powerful models; they need to be fed with high-quality, context-aware data that reflects real-world expertise. The 'experience transfer' concept is particularly interesting as it suggests a future where human skill can be digitized and shared more effectively than ever before. However, the challenge remains in how to accurately capture the 'why' behind human actions, which is often intuitive and hard to formalize. This will likely be a key area of research and development in the coming years.