대학생의 메타인지 수준에 따른 생성형 AI 질문 프롬프트 생성 및 활용 경험 연구

초록

The purpose of this study is to empirically examine the relationship between learners’ metacognition, prompt writing, and AI utilization experience in a generative AI-based learning environment. Forty undergraduate students from University A in Seoul participated in generative AI (ChatGPT)-based learning activities over four weeks, during which 4,591 learner-generated questions were collected, along with qualitative data on AI usage experiences. A mixed-methods design was employed, integrating quantitative and qualitative analyses. Regression analysis examined prompt generation patterns (number and length of questions), while multinomial logistic regression analyzed differences in question types according to metacognitive levels. In the qualitative phase, content analysis explored how AI usage experiences varied by metacognition. Results showed that higher metacognitive levels were associated with greater prompt frequency and length, as well as a higher likelihood of generating questions focused on critical thinking and evaluation rather than information search. Qualitative findings further indicated that learners with higher metacognition produced more logical, reflective, and higher-order questions and actively expanded AI responses, whereas those with lower metacognition primarily focused on information retrieval or prompt refinement. These findings suggest that learners’ metacognition is closely related to prompt utilization patterns and the qualitative depth of generative AI learning experiences, highlighting the need for instructional designs that incorporate metacognitive characteristics to enhance strategic question-generation skills.

키워드