South Korea explores AI-era overhaul of national statistics

Experts from government, industry and academia gathered Thursday to discuss ways to raise the value of national statistics and data in the AI era and to strengthen data quality management systems. Sessions focused on securing the data quality that drives AI performance and on applying AI to reform national statistics.

The Ministry of Statistics held the 2026 National Data Research Symposium at the Statistics Center in Daejeon on Thursday, under the theme "New Value and Use of National Statistics and Data in the AI Era."

The symposium, held annually since 2011, expanded its scope last year — when the National Data Research Institute was launched — from statistical methodology to data and AI more broadly. This year's event centered on data quality and integration, and on how AI can be used to modernize national statistics, amid what organizers described as a sweeping AI transformation.

In the keynote address, Ko Gil-gon, a professor at Seoul National University's Graduate School of Public Administration, spoke on the co-evolution of AI training data and statistical inference. He stressed the importance of national data in an era of AI-driven statistical reasoning and the need to shift toward training-ready datasets, and outlined a future direction and research agenda for the National Data Research Institute.

The symposium was organized into three sessions covering AI and data quality, AI-driven statistical innovation, and data integration and privacy protection.

The first session examined data quality management as a key factor in AI performance. The National Information Society Agency presented strategies for building large-scale, high-quality training datasets and securing quality in unstructured data, while the National Data Research Institute highlighted global research institutions' data quality management practices and stressed the importance of building and managing AI datasets.

The second session introduced AI-based statistical production cases, including the use of large language models for surveys and predicting infectious disease mortality. Proposals included using AI to improve survey data quality and automating quality management in national statistics.

The final session addressed data integration frameworks and privacy-protection technology. Researchers called for the introduction of data integration processes to improve linkage and usability, and for the development of new technologies to protect data used in AI training.

Minister An Hyeong-jun said "AI performance is determined by data quality and reliability," adding that "national statistics and data built accurately and systematically are core public infrastructure in the AI era." He said the ministry would continue to expand research aimed at improving data quality and creating new data value, drawing on the range of ideas presented at the symposium.