Generative Teaching via Code

arXiv:2601.04204v1 Announce Type: new
Abstract: The scalability of high-quality online education is hindered by the high costs and slow cycles of labor-intensive manual content creation. Despite advancements in video generation, current approaches often fail to ensure pedagogical structure and precise control due to their pixel-level, black-box nature. In this paper, we propose Generative Teaching, a novel paradigm that transitions educators from manual creators to high-level directors, allowing them to focus on pedagogical intent while autonomous agents handle the execution. To realize this vision, we introduce TeachMaster, a multi-agent framework that leverages code as an intermediate semantic medium. Unlike traditional video generation methods, TeachMaster orchestrates a collaborative team of agents–spanning planning, design, and rendering–to automate the production of interpretable, editable, and curriculum-ready educational videos. Experiments validate that TeachMaster significantly boosts production efficiency without compromising structural coherence or visual fidelity, providing a robust solution for scalable education.