The Korea Local Administration Review

This research aims to develop an automated model using generative AI to perform more efficient and consistent legal affairs investigations, with the goal of exploring the applicability of automated models to legal affairs surveys and charting future directions. The study presents two models: one for determining the nature of legal affairs (AI Model 1) and another for classifying types of legal affairs (AI Model 2). Through the application of these models, it was found that approximately 62% of efforts can be significantly reduced by eliminating non-administrative tasks, leaving around 320,000 tasks out of approximately 840,000 for review. Additionally, the role of automated models in verifying errors in existing data using trained artificial intelligence was evident.
Overall, the results suggest that legal affairs surveys, which previously required substantial effort, can be performed more swiftly and accurately through automated models. However, achieving this requires the development of more sophisticated models capable of reflecting various complex types of tasks. Until such advanced models are developed, exploring cooperative roles between humans (researchers) and AI will be essential.