공학 설계 활동에서 초등 영재의 생성형 AI 의존성 탐색

Abstract

This study aims to explore students’ reliance on generative artificial intelligence(generative AI reliance) in a STEM-based engineering design program. A total of nine gifted elementary students in grades 5 and 6 participated in a floating house engineering design activity supported by the generative AI tool Gemini. Data sources included students’ AI interaction logs, worksheets, prototypes, and survey responses. The analysis was conducted based on the core processes of engineering design: defining the problem, developing design solutions, and optimization. Students’ generative AI reliance in each process was operationally defined using a four-level scale (0–3). Both within-case and cross-case comparisons were employed to examine how generative AI reliance patterns emerged across the processes. Findings indicate that students’ generative AI reliance varied across the processes: higher reliance was observed in material selection and solution design, while more autonomous judgment and peer interaction were emphasized during the optimization process. These results highlight both the potential and the limitations of generative AI in STEM education, underscoring the need for clear pedagogical guidelines that balance AI support with students’ critical thinking and creative engagement.