SRL 기반 인간-AI 협업 루브릭 개발 모형 연구: 문헌사례연구를 중심으로

초록

This study aims to analyze how humans and AI collaborate in the process of rubric development using generative AI and to propose an SRL-based human-AI collaborative rubric development model. To this end, the study adopts the cyclical principles of self-regulated learning (SRL), the Human–AI–Human (H–AI–H) structure, and the OECD AI literacy framework as its theoretical lens. A literature case study method was employed, and seven studies related to AI-assisted rubric development and assessment design were selected from the literature collected through the Scopus and EBSCO databases. The analysis consisted of within-case and cross-case analyses, focusing on four dimensions: human pre-design and goal setting, AI-mediated generation and co-construction, human monitoring and refinement, and reflective and accountable redesign. The findings revealed three types of human-AI collaborative rubric development: the design-refinement cyclical type, the generation- adjustment-centered type, and the application-feedback support type. Based on these findings, this study proposes the DCA model, which consists of Designing, Co-creating, and Accountability. The DCA model demonstrates that AI should not replace the entire process of rubric development but should function as a collaborative mediator that supports assessment design between human goal setting and reflective judgment. This study is significant in that it conceptualizes generative AI-based rubric development not as an automated process of output generation, but as a human-centered and reflective assessment design process that integrates instructors’ AI literacy and responsible evaluative judgment.

키워드