Abstract: DeepSeek has sparked a new round of attention to generative AI, and set off a boom in the application of artificial intelligence in the field of education. For ideological and political courses, DeepSeek can promote the innovative development of ideological and political courses from three dimensions: "teaching", "learning" and "evaluation". In terms of "teaching", DeepSeek can help the systematical aggregation of teaching content, greatly improving teachers' efficiency in lesson preparation. In terms of "learning", DeepSeek can accurately recognize students' personalized needs and generate teaching content in response. Used in "evaluation", DeepSeek can effectively improve the accuracy and authenticity of ideological and political course teaching evaluations. However, DeepSeek-like generative artificial intelligence has its limitations in empowering ideological and political courses: "AI reliance" of educate participants tends to result in the loss of the meaning of education; "knowledge falsification" in the distortion of teaching content and ideological risk, and "discriminating data" in inadequacy in data utilization and risks of information security. Therefore, it is necessary to reshape the subject-object literacy of "human-AI integration", to promote the construction of artificial intelligence corpora, to optimize data governance effectiveness to enhance the practice path of DeepSeek-like generative artificial intelligence in empowering the ideological and political courses.

Key words: large model technology, artificial intelligence, ideological and political course construction, DeepSeek