Developing vibe-coding-based web tools for mathematics instruction and microteaching: Evidence-based analysis of learner participation and understanding by preservice mathematics teachers

This study explored how preservice mathematics teachers interpreted evidence of students' participation and understanding when using "vibe-coding"-based web tools during microteaching, and how they articulated improvement directions for tool-mediated lessons.Twenty-two second-year preservice mathematics teachers in Seoul participated in a 15-week course involving a vibe-coding lecture, web-based instructional tool development, a workshop with one-to-one feedback, revision, and microteaching.Final reflection journals were thematically analyzed, while lesson plans and artifact reports were used for contextual triangulation.The findings showed that preservice teachers tended to describe students' "high participation" mainly in terms of behavioral evidence such as submission rates, whereas evidence of discourse participation, such as spontaneous questions or student-led discussion, was limited.Judgments about understanding were mixed across quantitative formative assessment results, qualitative impressions, and descriptions in which no data were presented.Meanwhile, developing webbased instructional tools through vibe coding was perceived to offer advantages in visualization and rapid prototyping, but also constraints related to technical stability, usability, and the possible weakening of procedural performance.This study proposes design principles for teacher education that reconceptualize generative AIbased tool development not merely as "making tools, " but as fostering evidence-based reflection and instructional orchestration.