How Pre-Service Elementary Teachers Develop Scientific Concepts in AI-Integrated Lesson Designs: Implications for Sustainable Teacher Education

As AI and digital tools become more widely adopted in school education, integrating them sustainably into teacher preparation has become a central concern for sustainable teacher education. This study examined how pre-service elementary teachers develop scientific concepts within AI-integrated lesson plans and how those patterns change within each case following teaching demonstrations and instructor feedback. Qualitative content analysis was conducted on twelve lesson plans—initial drafts and revised versions from six groups across two science units—produced within an elementary science methods course. Plans were analyzed along three dimensions of conceptual development (conceptual structuring, generalization, and conceptual explicitness) and three functional roles of AI and digital tools. In draft plans, tools were predominantly used for learner engagement and artifact production, with scientific concepts embedded in activity contexts. Following feedback, conceptual explicitness was the dimension most frequently revised, while changes in conceptual structuring and generalization appeared in fewer cases. Cases in which conceptual development reached higher levels in revised plans shared a common design feature: AI outputs were repositioned within the consolidation stage in connection with explicit concept statements, rather than serving as content presentation. These findings suggest that pedagogical judgment about positioning AI outputs within lesson stages, reflected across design–demonstration–feedback–revision cycles, is central to the quality of AI-integrated science lesson design and offers implications for sustaining teacher preparation in the era of AI.