This study aims to explore elementary students’ developmental process of AI literacy through iterative cycles of design-based research (DBR) and to derive design principles for AI curriculum. A DBR approach was applied to 28 fifth-grade students at a South Korean elementary school, conducting three iterative cycles over 10 weeks. Each cycle consisted of four phases: analysis, design, implementation, and reflection. Data sources included AI attitude surveys, classroom observations, teacher interviews, and student journals. Findings revealed that, first, students progressed from anthropomorphic interpretations of AI to rule- and data-based understandings, and ultimately to critical thinking as designers. Second, across cycles, students showed significant increases in interest, ethical awareness, and perceived usefulness, while tension decreased. Third, ethical reasoning emerged through collaborative design and peer interaction without explicit instruction. Fourth, teachers’ adaptive scaffolding and collaborative knowledge construction were essential for addressing AI complexity. The study proposes four design principles for elementary AI curriculum: (1) leveraging anthropomorphism as a starting point for conceptual change, (2) gradual sequencing from concrete experiences to abstract principles, (3) experiential understanding through hands-on design, and (4) providing collaborative reflection experiences and pedagogical scaffolding.
You et al. (Sun,) studied this question.