Abstract: This article aims to address the teaching difficulties in the physics experiment course at Shanghai Jiao Tong University. Using artificial intelligence technology, based on the existing datasets in the experimental center, including lecture notes, lesson plans, courseware, exercise sets, and experimental reports, and using the open-source LlaMA2-7B model, a generative artificial intelligence model called Chat Physics education (ChatPED) specifically designed to assist students in their learning in the field of physics experiments is developed. And it was released for use by students during the course implementation phase, and a controlled experiment was conducted. The experimental group students used the model to assist learning, while the control group continued to attend classes as usual. Subsequently, the model was tested for its ability to improve students learning motivation and grades. Through a questionnaire survey of students (N=200), it was found that the experimental group students generally believed (75%) that the model was helpful for their learning, and the vast majority of students (90%) hoped to continue using it in other courses. Constructed a quantitative analysis of learning motivation, situational interest, and participation measurement scale. The Cronbachs alpha coefficient of the scale ranges from α=0.75 to α=0.81, and a pre/post use t-test was performed. The results showed that the learning motivation, situational interest, and participation measures of the experimental group students significantly improved after use compared to before use. Finally, double-blind grading was conducted, and the experimental group students improved their final scores by 0.2 points compared to the control group students (out of a total of 10 points). The significance P-test result is P<0.01, indicating that the improvement in academic performance of the experimental group students is related to the use of the model. Our research provides case studies of the application of artificial intelligence technology in specific teaching scenarios, which has certain practical significance.

Key words: University Physics experiment, artificial intelligence, large model, teaching