Abstract:
Artificial intelligence technology provides a new solution to the long-standing difficulties in teaching and learning the course of Theoretical Mechanics. Focusing on the systematic practice of AI-assisted teaching, this paper takes the general-purpose large language model DeepSeek as the research tool and explores two dimensions: problem-solving assistance and dynamic visualization. In terms of problem-solving, AI is employed to analyze the solution approaches to typical mechanics problems and, combined with manual correction, significantly improves the efficiency of teachers' responses to students' questions. In terms of visualization, AI is used to generate HTML5-based interactive animations, which to some extent addresses the problems of traditional animation resources, namely fixed content, lack of interactivity and high production costs. The practical constraints imposed on the effectiveness of AI-assisted teaching by differences in natural-language prompting are also discussed. Teaching practice shows that the AI-assisted class outperforms the control classes in conceptual understanding, final examination scores and classroom atmosphere. AI technology not only reduces teachers' repetitive workload but also helps lower the threshold for producing high-quality interactive resources, making it possible to narrow resource gaps and providing a replicable practical paradigm for the digital transformation of the Theoretical Mechanics course.