AI-aided-RIA(AI를 활용한 규제영향분석)의 가능성 탐색: 산업안전보건규제사례를 중심으로

In this study, problems occurring in the regulatory impact analysis process were theoretically confirmed. It was suggested that the use of AI can overcome these problems. Furthermore, it exemplified that AI can overcome problems in the implementation process through the case of occupational safety and health regulation regulatory impact analysis. First, the general difficulties of regulatory impact analysis were analyzed with three theories of Rational decision-making theory(analysis), Principal-agent theory(control), and Governance theory(participation). It presented how the difficulties of each regulatory impact analysis could be overcome if AI was introduced. Afterwards, based on the case of Occupational Safety and Health Regulatory RIA, the essential difficulties in the process of conducting the regulatory impact analysis were divided into four areas: identification of regulatory impact groups, deriving scenarios, and evaluation and deduction of costs and benefits of regulatory impacts. AI techniques and tasks that can be specifically used in each of the four fields are exemplified. As a result, in the analytical aspect, AI possesses the strengths of data collection and organization, powerful computation, and optimization computation, so it is possible to ① time and budget saving, ② analysis technique improvement and objectification of subjective value, ③ mass data handling, ④ securing timeliness, and ⑤ Conduct comparative analysis of alternatives. In terms of control, continuous monitoring made it possible to ① identify the owner's preferences, ② prevent capture and become independent from political logic, and ③ enhance expertise and acquire information. Lastly, in the participatory aspect, it was possible to collect and reflect real-time public opinion through the expansion and management of its own data. Therefore, by introducing AI into specific administrative areas (regulatory impact analysis), evidence-based administration centered on objective data is possible. In addition, through participation and two-way communication, AI can assist in the overall institutional decision-making process.