Dual-contrast pedagogy for AI literacy in upper elementary schools

Advances in artificial intelligence (AI) have highlighted the need to equip young students with basic AI-related knowledge, skills, values, and attitudes. However, pedagogical design for AI literacy remains a critical challenge, especially for upper elementary students aged 10–12. This design-based study had two goals: to develop a pedagogical approach for AI literacy in upper elementary education and to empirically assess this approach through an experiment. One hundred forty-seven sixth graders in an upper elementary school were randomly assigned to a control group (n = 75) and an experimental group (n = 72). Following a theory-informed design convention, we proposed a dual-contrast pedagogical (DCP) approach. This approach centers on human-AI comparisons by integrating analogies and cognitive conflicts. Two teaching examples on machine learning and large language models were provided. The experimental group was taught with the DCP approach, while the control group received conventional direct instruction. Data drawn from assessment tasks and questionnaires were subjected to two-way analyses of variance and covariance. The experimental group demonstrated significantly higher performance in AI knowledge, skills, and ethical awareness. They also exhibited a significant increase in AI learning confidence and intrinsic motivation and a significant decrease in learning anxiety. The DCP approach significantly improved students' learning performance and attitudes, demonstrating its effectiveness in promoting AI literacy. This study highlights the pedagogical value of human-AI comparisons in teaching AI, while contributing to a research agenda on the cognitive and conceptual aspects of AI education.