생성형 AI 기반 대학창업교육에서 인간-AI-동료 협업과 메타인지적 조절에 관한 탐색적 사례연구: 서술형 성찰 자료에 대한 텍스트마이닝과 질적 내용분석의 통합 접근

초록

This exploratory study examines how 39 undergraduates in a single entrepreneurship course described their collaboration with generative AI and peers, and how they narrated regulation of their own thinking during that process. Using 390 response units (10,666 tokens) from open-ended final-exam items, text mining and semantic network analysis identified structural patterns, followed by qualitative content analysis for interpretation. Six themes emerged: students framed generative AI not as a search tool but as a thinking partner for divergence, structuring, verbalization, and perspective shifting. Encountered uncertainty—hallucination, genericness, contextual limits—prompted question refinement and critical review, narrated as metacognitive regulation. Peers supplied market realism, differentiation, value judgment, and self-understanding that AI did not provide. In the semantic network, “question” showed maximum degree centrality, while peer-related terms remained peripheral—an asymmetry the qualitative strand helped explain. As the data are self-reports from one course under assessment conditions, findings are bounded to perceived rather than measured collaboration or metacognitive change. The study conceptualizes generative AI as an asymmetric quasi-node in perceived human-AI-peer collaboration, positions metacognitive regulation as a perceived linking mechanism, and demonstrates a structure-first sequential mixed-methods design for exploratory research in educational IT.

키워드