"Korea is a blessed country": The future industry that global Big Tech is eyeing (Mobiltech CEO Kim Jae-seung)
- What to look for
- Viewers can learn how physical AI is trained and evaluated using digital twins, sensor calibration, and local data before entering the real world, and identify differences between the infrastructure needed at the research stage and for large-scale operations.
- Why it matters
- This helps viewers examine the robotics industry in terms of data, validation, and maintenance as well as models and hardware, and separately assess experimental environments, real-world performance, safety, and data protection conditions behind companies' commercialization claims.
AI summary
Mobiltech's CEO explains that commercializing physical AI requires not only robots and models but also training and evaluation environments incorporating real-world data, along with sensor calibration. His view is that digital twins can replicate roads, factories, and logistics spaces to enable repeated testing of hazardous situations, but measuring and reducing the gap between reality and simulation is important. He explains that end-to-end AI is affected by its environment and context, so local data and retraining are necessary for it to work well in Korea. He regards Korea's diverse manufacturing sites and demand for actual work as assets, arguing that collecting, managing, and standardizing data can contribute to industrial competitiveness. He introduces his company's strategy of pursuing business opportunities in data, simulation, and validation infrastructure supplied to multiple developers rather than in a single finished product. He explains that moving from a small number of research devices to large-scale operations makes production, calibration, maintenance, and reliability management important, while safety and autonomy remain major challenges for household robots. Data from workplaces and homes involves both permission and privacy issues. Statements about the company's technological edge, market outlook, and regulatory changes should be distinguished as the executive's account, and simulation results do not guarantee real-world safety. All directional outlooks are the speaker's conditional judgments, not definitive short-term price predictions.