AI 시스템의 위험 완화를 위한 정책적 접근방안 연구: AI 영향평가를 중심으로

This study aims to conduct an exploratory case analysis of AIA(AI Impact Assessment) published in the public sector as a policy approach to mitigating the risk of AI systems. This study constructs a research analysis framework based on the key points of AIA design presented in previous studies. Based on the subsequent analysis framework, six AIA cases implemented or proposed as policies by national governments or public institutions to date have been adopted and analyzed. As a result of the analysis, there was a perception that AIA should recognize technical and industrial specificity different from the existing impact assessment, and the AI system was comprehensively and broadly defined. Although some opinions were gathered regarding risk mitigation measures, the way the mitigation measures presented in the AI impact assessment were linked to the level of risk classification differed from country to country. In addition, it was carried out regularly at various points in the AI life cycle and had a strong self-regulatory character. This study contributes to a deeper understanding of AIA development by presenting comprehensive data on policy and research trends, thereby informing future policy development in this field.