Integrating generative artificial intelligence into secondary school programming education: model design and empirical evidence

Programming education is essential but remains challenging in secondary school contexts under traditional instructional approaches, which often struggle with the high cognitive load, debugging difficulties, and lack of personalised feedback. To address these, this study designed a GAI-supported programming teaching model for secondary school programming instruction, structured into four phases: introduction, knowledge presentation, coding practice, summary and reflection. Meanwhile, this model delineates the specific responsibilities of students, teachers, and GAI at every phase. To evaluate the effectiveness of this model, a quasi-experimental study was conducted in a high school in East China, involving 100 first-year students in a three-week Python course, where the experimental group adopted the GAI-supported programming teaching model (n = 52), and the control group followed the traditional programming teaching model (n = 48). Results showed that the experimental group achieved significantly higher programming performance, learning satisfaction, and programming self-efficacy, as well as improvements in computational thinking, particularly in algorithmic thinking and problem-solving. However, no significant improvement was found in creativity and critical thinking. The findings highlight the effectiveness of integrating GAI into secondary-level programming teaching and offer practical insights for educators to design GAI-enhanced learning environments that support personal ability development.