Supporting Feedback, Not Replacing It: University Supervisors’ Perceptions of Generative AI in Preservice Teacher Field Experiences

Artificial intelligence is increasingly shaping educational practices; however, its use by university supervisors (USs) to provide feedback to preservice teachers (PSTs) during field experiences remains underexplored. This collaborative action research study examined USs’ perceptions of implementing a generative AI (GenAI)-supported feedback protocol during classroom observations. Supervisors used a researcher-developed protocol in which observational notes, lesson objectives, and teaching competencies were input into ChatGPT to generate draft feedback for post-observation conferences. Data sources included annotated protocol documents, individual interviews, and a focus group, supplemented by existing PST interview data. Inductive thematic analysis indicated that GenAI supported alignment with teaching competencies and enhanced the structure and specificity of feedback. At the same time, findings highlighted important limitations, as AI-assisted feedback required careful human interpretation to ensure contextual accuracy and relevance. Supervisors noted that, while GenAI provided objective-aligned instructional guidance, it did not fully capture the complexity of classroom interactions. These findings suggest that GenAI functions as a support tool rather than an autonomous feedback mechanism, underscoring the importance of human judgment in AI-assisted supervisory feedback within teacher education.