자문화기술지 분석을 통한 인간-AI 협업 체계적 문헌고찰 방법론의 비판적 탐색

초록

This study explores the structure, applicability, and epistemological implications of the Human-AI Collaborative Systematic Literature Review (HAC-SLR) methodology. While artificial intelligence (AI) is increasingly used for literature review tasks, a methodology that systematically integrates human and AI roles throughout the entire review process remains underdeveloped. An instrumental case study was conducted by applying HAC-SLR to 30 Korean autoethnographic studies on healing experiences published between 2015 and 2025. Data included prompt records, AI outputs with human judgments, role-adjustment records, and researchers’ reflective journals. Inductive categorization and constant comparative analysis were employed. The findings identified six stages of HAC-SLR: research question formulation, literature search, study screening, quality appraisal, data analysis and synthesis, and abstraction and interpretation. Four methodological principles emerged: stage-specific role differentiation, human authority over final decisions, iterative refinement through AI-assisted analysis and human review, and transparent documentation through a four-layer recording system. The application of HAC-SLR conceptualized healing as a four-stage process-existential rupture, structuring of affect, narrative reconstruction, and relational reconfiguration-and revealed three epistemological tensions: the pre-structuring effect of AI outputs, the tension between autoethnography and AI-based pattern detection, and the researcher’s triple positionality as designer, practitioner, and analyst. This study extends AI-assisted systematic literature reviews beyond literature screening toward collaborative qualitative synthesis and theory construction, offering a methodological framework for ensuring transparency, accountability, and epistemological rigor in human-AI collaborative research.

키워드