Abstract: To address the challenges of abstraction and conceptual comprehension in college physics teaching, this study proposes a generative AIbased instructional program design method. Using the diffraction grating spectral with monochromatic light visualization as a case study, a webbased adjustable visualization platform was developed through a "teacherAIstudent" collaborative system. The research adopts an iterative process involving AIassisted code generation, teacher verification, and interface optimization to achieve lowcode development of teaching demonstrations. Teaching practice with 48 valid questionnaires showed that 91.7% of students agreed that the tool helped them understand abstract concepts, 95.8% reported enhanced learning interest, and 93.8% were willing to continue using it after class. The findings indicate that the AI system, functioning as an intelligent scaffold, effectively reduces cognitive load and facilitates the cognitive transition from phenomenological understanding to abstract reasoning, providing new pathways for physics teaching reform.

Key words: generative AI, college physics, constructivism, dynamic visualization, low code teaching development