인간–AI 협업 질적연구를 위한 IRB 프로토콜 개발에 관한 연구

초록

This study aims to develop an Institutional Review Board (IRB) protocol for humanAI collaborative qualitative research, seeking to ensure the interpretive responsibility of human researchers and the protection of research participants as generative Artificial Intelligence(AI) becomes increasingly embedded in qualitative research. AI is now used in key research stages, including text summarization, categorization, and initial coding, thereby transforming conventional research practices. However, the Belmont Report’s ethical principles alone cannot fully address emerging concerns such as data re-identification, algorithmic bias, and unclear responsibility attribution. To respond to these issues, this study conducted a benchmarking analysis of international ethical and governance frameworks, including NIST, GDPR, and UNESCO, and administered an expert survey from ten specialists in qualitative research, research ethics, and artificial intelligence. The findings identify five core principles: transparency in AI use, risk management with data protection and re-identification, expansion of informed consent, control of algorithmic bias, and ultimate accountability of human researchers. Strong consensus emerged on the necessity to define human accountability and AI usage procedures clearly. Accordingly, this study proposes shifting IRB review’s focus from evaluating AI systems’ technical properties to examining human researchers’ procedural safeguards. The protocol includes documentation of prompt design history, tracking of AI tool’s specification, cross-validation procedures, and data disposal frameworks. This study contributes to redefining qualitative research standards in the AI era, enhancing the effectiveness of IRB reviews, and protecting research participants’ data sovereignty.

키워드