#1102: Shared Epistemic Agency in K-12 AI-Infused Lesson Planning: Exploring Human-AI Collaboration in Designing Pedagogically Meaningful Learning Experiences

This study examines how K-12 teachers engage in shared epistemic agency with generative AI tools during the co-construction of AI-integrated lesson plans. Using a mixed-methods design, we analyzed data from 36 in-service teachers enrolled in a graduate-level course. Data sources included 36 lesson plans, and 332 teacher-AI interaction episodes across 68 chatlogs. Rubric-based evaluation scored instructional quality, GenAI integration, and epistemic agency, revealing strong positive correlations among these dimensions. Findings show that teachers demonstrated strong epistemic agency in AI collaboration, actively framing tasks, evaluating AI outputs, and maintaining human control through the guidance of epistemic aims, ideals, dispositions and epistemic metacognitive strategies. Shared epistemic agency emerged when teachers and students co-regulated inquiry and epistemic control was distributed, regulated, and reflected upon across human-AI turns. Qualitative analysis guided by the Epistemic Conjecture Framework (Barzilai & Chinn, 2024; Sandoval, 2014) showed that lessons integrating AI through purposeful tools, activity structures, and discourse supported reliable epistemic processes demonstrating that well-designed AI use can cultivate epistemic growth.