Abstract
This study aims to propose an educational intervention to promote pre-service teachers’ creative metacognition in the context of teacher education by developing an LLM-based educational conversational AI agent and exploring its applicability. Creative metacognition refers to a higher-order cognitive system through which individuals monitor and control their thinking throughout the creative problem-solving process. In this study, particular attention was given to question-centered scaffolding that supports such metacognitive processes across different stages. Accordingly, the developed AI agent was designed to elicit metacognitive thinking in alignment with the stages of creative problem solving. An analysis of pre- and post-changes in creative metacognition among 25 pre-service teachers showed that overall levels of creative metacognition significantly improved, with particularly meaningful effects found in the stages of problem understanding and idea generation. In addition, sequential pattern analysis of learner-agent interaction logs revealed recurring patterns of metacognitive monitoring and control. These findings suggest that question-centered scaffolding delivered through an LLM-based conversational AI agent can foster pre-service teachers’ creative metacognition and provide structural support for the creative problem-solving process. Furthermore, this study offers implications for designing and utilizing LLM-based conversational AI agents as scaffolding tools for creative metacognition in teacher education.