JEJU — Park Young-sun, chair of the strategic economy advisory group at the Ministry of Finance and Economy, said smaller manufacturers should build AI systems around industry-specific cooperatives and overhaul their internal organizations to make the shift to physical artificial intelligence. She also called for a Korean data standard, which she termed a "K-ontology," to link and make effective use of the data companies already hold.
"The outcome of the AI race will ultimately be decided by data," Park said in a keynote address on the 21st at a policy forum on innovation strategies for smaller firms in the physical AI era, held at the Lotte Hotel Jeju and co-hosted by the Korea Federation of SMEs and the Ministry of Finance and Economy. "What matters is connecting the data that smaller companies hold through industry cooperatives and turning it into an asset, in a form AI can understand."
Park said companies should first adopt agentic AI and reorganize themselves before moving to physical AI. "Before bringing in physical AI, companies have to become organizations that work alongside AI and change how things are done on the shop floor," she said. "Chief executives need to use today's agentic AI themselves first, and then redesign their organizations and work processes on that basis." Under that approach, people would handle the critical tasks such as setting objectives and verifying results, while AI takes on the intermediate steps.
She also stressed the importance of ontology, which allows AI to grasp the meaning of data and the relationships within it. "Ontology is what links different sets of data so that AI can understand the relationships and the context between them," Park said. "Physical AI can only succeed here if the country builds this kind of K-ontology."
Park further argued for the need to build AI factories — intelligent production platforms that learn from and analyze manufacturing data to support the operation of production equipment and robots. "Going forward, every company will need two factories," she said. "One is the manufacturing plant that makes products, and the other is the AI plant that makes artificial intelligence." She added: "Since smaller manufacturers cannot build this on their own, cooperatives should take the lead and the Korea Federation of SMEs should serve as the hub that connects them."
Park also voiced concern about the government's sovereign AI foundation model project. "I am somewhat worried about the sovereign foundation model the Ministry of Science and ICT is pursuing," she said. "If it goes wrong, there is a chance it ends up like Hangul and Computer, used only within the country." She was critical of the data strategies at NAVER (035420) and Kakao (035720), the country's leading platform companies. "Naver accumulated knowledge but failed to connect it," Park said. "Kakao holds relationship data on the entire population of South Korea, and had it used that to build its own AI agent, it could have become a global company. It did not."
In presentations that followed Park's keynote, speakers also stressed turning manufacturing data into an asset and building shared infrastructure. Kim Jun-ha, dean of the AI Policy and Strategy Graduate School at the Gwangju Institute of Science and Technology, said smaller companies must go beyond simply collecting data. "They have to turn it into an asset that AI can use," he said. "Cooperatives should set common standards and provide a shared basis for testing, while the state supplies the infrastructure to support that — a division of roles is needed." Cho Joo-hyun, president of the Korea Institute for Small Business Policy and Planning, said a "joint innovation model" is needed in which common foundations such as data and testing facilities are shared while individual companies retain their own competitiveness. "A cooperative framework is also needed to spread the successes of leading companies quickly to other firms in the same industry," he said.