[ROUNDTABLE] Korea needs AI-ready ontology, should not end with developing models

The roundtable highlighted a strategic shift for Korea's artificial intelligence development, moving beyond its reported focus on advanced model creation. Experts called for prioritizing "AI-ready ontologies," defined as structured knowledge systems designed to organize vast government data and institutional knowledge into formats AI systems can comprehend and use. This approach suggests that effective AI deployment requires not just powerful models, but also the underlying data infrastructure to feed them meaningful, organized information.

The practical value of these ontologies was demonstrated through specific examples. Reform Party of Korea Rep. Lee Jun-seok's office converted raw data from the Public Procurement Service into ontologies, then used an AI program to analyze it. This analysis reportedly uncovered signs of potential misuse of 562.8 billion won ($418.4 million) in taxpayer money. An ongoing project to convert Korea's laws into ontologies has similarly identified "broken links between legal provisions," illustrating how structured knowledge systems can reveal systemic issues.

Beyond ontologies, the discussion also touched on broader strategic elements. Rep. Lee Jun-seok advocated for a purpose-built AI city in Saemangeum to facilitate testing under flexible regulations. Concurrently, Lim Jong-in emphasized the importance of Korea maintaining its own foundation models for sensitive applications like national security, chip design, and cybersecurity, citing Samsung Electronics' use of its internal Gauss model alongside foreign-developed AI to protect proprietary information.