How Do Critical Evaluations of Generative AI Outputs Shape Teacher Agency? A Mixed-Methods Comparative Analysis

Generative AI (GenAI) promises educators rapid creation of instructional materials through simple prompting, but it may also shift teachers from active pedagogical designers to passive consumers of AI-generated outputs. This raises concerns about the potential erosion of pedagogical integrity and teacher agency. To address this issue, this study investigates how teachers can reclaim agency in learning design by critically evaluating and shaping GenAI outputs. Drawing on the Triadic Reciprocal Framework of Core Agency Concepts (TRFCAC), we analyzed teachers’ in-the-wild interaction traces and post-task reflections as they co-designed course materials with GenAI. The findings show that GenAI integration did not necessarily boost process efficiency for critical teachers in creating learning materials, whereas non-critical teachers often relied on GenAI outputs without attending to potential risks, highlighting a potential tradeoff between task efficiency and teaching quality.