Yann LeCun Says Physical AI Could Give Korea a Global Edge
AI pioneer Yann LeCun says South Korea can turn its manufacturing, robotics and engineering base into an advantage as artificial intelligence moves beyond language models and into the physical world.
SEOUL — South Korea could gain an important advantage in the next phase of artificial intelligence by applying the technology to factories, ships, machinery and robots, AI pioneer Yann LeCun told an industrial AI gathering in Seoul on Monday.
LeCun said the next major advance is likely to come from systems that can understand, predict and act in the physical world rather than from simply making large language models bigger. He described so-called world models as systems capable of anticipating how the environment will change in response to an action.
That approach could allow an AI system to plan a sequence of steps needed to reach a goal, instead of only predicting the next word or reproducing patterns found in human-created data. LeCun, a 2018 Turing Award recipient and the founder and chairperson of AMI Labs, is developing the concept through a joint-embedding predictive architecture.
The shift could suit Korea because of its established strengths in semiconductors, automobiles, shipbuilding, batteries, robotics and industrial engineering. Speakers at the panel argued that Korea does not need to compete only by matching the vast computing investments of major U.S. technology companies. It could instead focus on deploying AI in sectors where it already possesses deep operating expertise and valuable industrial data.
Aitomatic CEO Christopher Nguyen said earlier waves of digital technology largely remained in cyberspace because they could not interact effectively with the physical world. AI is now beginning to carry digital intelligence into factories and machinery, he said, urging Korea to deploy industrial systems quickly while strengthening security and governance.
Jimmy Rhee, chairman of ASI Foundry, characterized the country’s industrial data as a strategic resource. Korea’s manufacturing base and research talent create an opening, he said, but waiting for perfect information could allow competitors to move ahead.
Ha Jung-woo, vice chairman of South Korea’s National AI Strategy Committee, said global competition is shifting from headline model performance toward industrial productivity. He said international cooperation will be important in converting Korea’s capabilities in chips, vehicles, shipbuilding and batteries into economic value.
LeCun also called for stronger efforts to attract global researchers and connect fundamental research with commercialization. As foundation models increasingly become open infrastructure, he said, Korea could build on them to seek leadership in physical AI and robotics.
Addressing recent AI-related attacks on Korea’s financial sector, LeCun rejected the view that they showed AI becoming uncontrollable. He argued that the incidents instead revealed weak cybersecurity safeguards, including inadequate expertise among companies deploying agents and software sandboxes that failed to contain them securely.