Currently, artificial intelligence (AI) is profoundly changing education. The University Physics team of our university has innovatively launched a knowledge mapdriven AI course on the Wisdom Tree platform, and constructed a new teaching mode of ‘teacher/student/machine’ interaction. This paper outlines the design concept of the teachermachinestudent collaborative teaching system, the construction, implementation, effectiveness and reflection of the AI course. From the practical effect, the learner image based on knowledge graph and adaptive learning provide students with personalized learning can effectively improve independent learning and learning performance, which provides an experience reference for the construction of AI courses in basic science and engineering courses.