Traditional university programming education follows a “theory-practice” model, aiming to cultivate students’ computational thinking. However, research has shown that teachers can only impart pseudo-code and use it for extremely simple cases, limiting the scope of practice. In response to this issue, a “practice-theory-practice” teaching model has been proposed. The research team developed two AI service tools, Cheng Xiaoming and Code Monkey, based on the artificial intelligence generation platform Prompt Sapper, which combine providing specific code examples (teaching people to fish) with offering methods for writing code (teaching people how to fish). Through designed experimental schemes, the traditional university programming education model was compared with the new AI-powered programming education model. The new teaching model, assisted by AI, has been validated in practical teaching applications to enhance programming skills and cultivate computational thinking, demonstrating its effectiveness and impact.