Sakuragi Junior High School, Kyoto, Japan

Abstract: Based on the policy framework of Japan’s Education DX Roadmap, this study targets core pain points in junior high mathematics teaching, including difficulties in understanding abstract concepts and insufficient personalized guidance. It proposes a dual-cycle human-machine collaborative teaching model for generative AI-enabled mathematics instruction. With empirical cases focusing on quadratic function teaching in Gifu prefectural junior high schools and geometric proof teaching in affiliated schools of Osaka University of Education, this research verifies the innovative practical value of generative AI in dynamic representation, cognitive conflict design, and hierarchical teaching guidance. The results indicate that the proposed model significantly improves students’ mathematical thinking depth (problem-solving ability increased by 26.4% in the experimental group) and learning engagement (sustained inquiry behaviors increased by 42%). In accordance with Japan’s educational data standard (DMP) requirements, this study constructs a three-layer implementation framework consisting of a technology adaptation layer, a teaching strategy layer, and an ethical norm layer, providing front-line teachers with a progressive development path from tool application to thinking reconstruction. This research confirms that teachers’ digital literacy and AI ethical judgment are the core determinants of sustainable technological application. The AI teaching maturity matrix proposed in this paper can provide practical references for the digital transformation of junior high mathematics education in Japan.

Keywords: generative AI; junior high mathematics; educational digitalization; teaching innovation; human-machine collaboration