Online and offline blended teaching has become the main form of physics teaching style, but the survey found that the online resources of blended teaching have problems that are difficult to solve, such as insufficient interactivity and innovation, learning effect and information differentiation, and low achievement of abstract knowledge comprehension. Based on the current artificial intelligence AI open source tools, this paper explores the use of AIassisted experiment design innovation tasks embedded in blended teaching, through the stepbystep design of independent exploration experiments, to realize the optimization of experimental methods and details by AI technology, promote students' active learning in the process of blended teaching, improve students' ability to realize AIassisted physical experiment design exploration based on personal learning characteristics, and enhance learners' physics situational experience. The results of the questionnaire survey show that the students' knowledge comprehension, memory and internalization, innovation and collaboration ability, and positive learning emotion have been significantly improved by AIassisted experimental physics scenario construction.