In the practice of interdisciplinary instructional design, teachers often encounter difficulties in aligning core competencies, insufficient integration of interdisciplinary knowledge, limited activity design, and weak implementation feasibility, which restricts the educational effectiveness. In recent years, the in-depth application of large language model (LLM) in the educational field is promoting the transformation of teaching designs towards intelligent collaboration, offering new ideas to solve the above problems. Based on this, the paper summarized the four applicability of LLM in supporting interdisciplinary instructional design (namely parameterized knowledge, world model, emergent capabilities, and theory of mind) and utilized this as the functional support to construct the framework for interdisciplinary instructional design supported by LLM. Subsequently, this paper selected the thematic practice class “gravitational potential energy” in high school physics as an application case. The teacher team carried out the teaching application followed the framework's four stages of the interdisciplinary goal system establishment, interdisciplinary teaching activity design, interdisciplinary resource retrieval and construction, and interdisciplinary teaching evaluation and optimization. Through expert scoring and interview transcript analysis, this paper found out that the experts' scoring of the interdisciplinary instructional design generated by the case reached a good level, and that the application of the framework could help to implement core competencies, deepen disciplinary integration, empower comprehensive activity design, and ensure practical feasibility. The research in this paper could provide theoretical guidance and practical insights for enabling LLM to empower interdisciplinary instructional design.

With their robust analytical and inferential capabilities, large language model (LLM) are transforming educational research paradigms, particularly made significant advancements in agent technology, which provides strong support for systematically solving complex problems in the scientific research field. Based on this, the paper focused on the typical research task scenarios of meta-synthesis, and discussed how to provide more systematic support with an agent-based approach. Firstly, this paper introduced the application principles for meta-synthesis agent application, including multi-step planning, collaborative mode construction, prompt empowerment, and tool integration, designed an application mode involving the coordinated efforts of six agents, as well as developed a meta-synthesis agent tool based on this mode. Then, the agent tool was applied to typical meta-synthesis tasks through case studies. It was found that compared to human teams, the agent can perform the task in accordance with the meta-synthesis research process better and generate more comprehensive results. Meanwhile, human teams gave positive evaluation on the accuracy and user experience of the agent during the application process. Finally, based on the research findings, this paper put forward the application strategy of the agents in educational research, in order to offer a new insight of man-machine collaboration for solving the practical problems of educational research.