From Auxiliary Tool to Teaching Partner: A Human-AI Co-teaching Model for Primary Programming Classrooms

AI chatbots are increasingly applied into programming education, they are now often positioned as auxiliary tools for student self-learning. However, for primary school students, effective programming learning requires instructional scaffolding. This study develops a human-AI co-teaching model that reframes AI chatbot as a teaching partner collaborating with human teachers in primary programming classrooms. Using action research method, the model was iteratively designed and implemented in two primary schools’ students (N = 75). In this model, the AI chatbot acts as a lead teacher by explaining programming concepts, guiding learning tasks, and providing immediate feedback through conversational interaction, while human teachers focus on classroom organization, regulation, and learner support. This model was evaluated in students’ cognitive outcomes (programming knowledge and skills) and non-cognitive factors (self-learning behavior, self-regulated learning and learning experience). Results indicate that students achieved satisfactory programming performance and significant improvements in self-regulated learning, particularly in learning motivation and process. These findings emphasize that effective programming education relies on the human-AI instructional design, and AI serves as a teaching partner rather than replacing human teachers.