Generative AI highlights the contrast between students’ dualistic epistemic practices and teacher education learning objectives

This study examines student teachers’ capabilities when adopting a generative AI system as a new cognitive tool. In our pedagogical intervention, students used ChatGPT 3.5 to support a small research task. Consistent with decades of research on higher education students’ epistemic positions, most students approached the knowledge-building task (and, respectively, ChatGPT) with dualistic epistemic practices. Notably, ChatGPT’s polished interface invites naïve dualistic interpretations. However, teacher education learning objectives and effective knowledge-building with generative AI tools require more sophisticated epistemic stances: understanding knowledge as contingent and context-bound and knowledge-building as an activity that requires validation. This suggests that the central challenge for teacher education is not generative AI per se but supporting students’ epistemic development so that they can use such tools responsibly.