摘要
The rapid advancement of artificial intelligence technologies has driven the widespread adoption of generative AI (GenAI) in programming education. Owing to its powerful capabilities in language understanding and generation, GenAI has shown considerable potential in supporting programming instruction. However, students often lack the experience and skills needed to ask effective questions, which makes it difficult for them to generate high-quality queries or to deeply interpret and utilize the feedback provided by GenAI. This limitation can hinder both the depth and effectiveness of programming learning. To address this issue, the present study proposed a GenAI-assisted programming learning approach that integrates Socratic questioning method, aiming to guide students in engaging more deeply with GenAI during programming tasks. A quasi-experimental design was employed, involving 69 primary school students aged 10 to 11 who participated in a nine-week intervention. The experimental group (n = 35) received GenAI-assisted programming with Socratic questioning, while the control group (n = 34) received GenAI-assisted programming instruction without Socratic questioning method. In this comparison of two intact classes, the experimental group demonstrated higher post-intervention scores in computational thinking and metacognitive awareness than the control group, while no statistically significant difference was observed in learning achievement. Given the study design, these findings should be interpreted as preliminary evidence that Socratic questioning may support the pedagogical use of GenAI in programming education. The results also suggest potential theoretical and pedagogical implications for designing AI-scaffolded learning environments that aim to foster higher-order thinking in K-12 programming contexts.