Companies and public institutions strengthen data cleanup to use AI

Companies and public institutions are accelerating efforts to organize documents and statistics into a format that AI can use directly in line with the AI transformation. This is because if there are many old files and duplicate documents, AI agents may have difficulty finding the information they need, reducing work efficiency. According to recent studies, when only the materials needed for AI were provided, information retrieval accuracy and task completion rates were higher than when all documents were handed over at once. This shows that, more than AI's reasoning ability, the process of selecting the information suited to the current task from among vast amounts of material can become a bigger bottleneck. This trend is not limited to companies internally. The United Nations and the World Bank are linking statistics and reorganizing data so that AI can access official data directly, and Korea has also launched a pilot service connecting the National Statistical Portal (KOSIS) with AI. Experts say that to make good use of AI, it is important not only to introduce models but also to reorganize data structures and access permissions together.