Abstract: Mechanics of Materials, as a core fundamental course for engineering majors, faces persistent challenges such as high abstraction, complex derivations, and low student engagement. Based on cognitive tool theory and the concept of human-machine collaboration, this paper constructs an AI-empowered teaching model integrating “cognitive co-construction, cross-modal reshaping, and process tracking feedback.” The model employs Generative AI to achieve intelligent transformation among language, formulas, images, and animations, establishing multidimensional cognitive channels while enabling knowledge co-construction, real-time feedback, and dynamic regulation throughout instruction. Empirical results from a two-year teaching practice demonstrate that this approach effectively reduces the failure rate and significantly improves the excellence rate. It enhances students' visualization and understanding of mechanical concepts as well as their engineering application capabilities. This study promotes a pedagogical shift from “knowledge inculcation” to “cognitive generation,” providing a new perspective and empirical evidence for the intelligent transformation of fundamental mechanics education.