Effect of generative AI-based conceptual and metacognitive scaffolding on elementary students’ computational thinking, academic achievement, and programming self-efficacy

Elementary students face challenges in comprehending computational thinking (CT) concepts and designing solutions. Scaffolds supported by generative AI (GAI) offer an effective approach for cultivating CT. This study employed a pre-post quasi-experimental design to investigate the different effects of GAI conceptual vs. metacognitive scaffolding on CT in elementary students’ visual programming. Results indicated that the GAI metacognitive scaffolding demonstrated significant advantages in enhancing CT and programming self-efficacy. While the GAI conceptual scaffolding promoted academic performance, this advantage was not significant. This study enriches empirical research on integrating GAI with elementary programming education, providing theoretical and practical guidance for designing personalized CT development for younger learners.