The purpose of this study is to compare and examine the results of discourse analysis in lessons conducted using the lesson-specific model developed by Hoshimori (2024) with the results of discourse analysis using a new general-purpose Large Language Model (LLM). The research method involves comparing the results of an analysis using the class-specific model trained on speech protocols labeled by Hosomi (2024) (an analysis using the class-specific model trained on speech protocols labeled by Hosomi (2024) from a mathematics class taught by an in-service elementary school teacher, focusing on “active, interactive, and deep learning”) with the results of an analysis using a new general-purpose LLM (Large Language Model). ), we conducted speech analysis using a new general-purpose Large Language Model (LLM) on the same mathematics lessons. We then compared the results from these two approaches. Through comparing the results of the lesson-specific model and the general-purpose model, we aim to develop an AI reflection tool that encourages active self-reflection among teachers and teacher candidates. In this presentation, we wish to discuss future prospects with participants based on the comparison results.
