Student Interaction with ChatGPT During a Source-Interpretation Task in Initial Teacher Education: Implications for Critical Thinking

Abstract

Social Sciences education requires pedagogical approaches that foster critical thinking and enable students to analyze, contrast, and interpret complex social realities. As Generative Artificial Intelligence (GenAI) becomes integrated into higher education, it offers opportunities to support these processes while raising questions about students’ autonomy and reliance on generated content. Although previous research has examined the educational uses of GenAI, less is known about how students interact with these systems when revising source-based interpretations and how they position themselves. This study examines how 83 first-year university students enrolled in Education degree programs interacted with ChatGPT during a source-interpretation activity and how this interaction was reflected in the reformulation of their initial responses. A qualitative descriptive–interpretative design compared initial responses, prompts, revised responses, and retrospective reflections. The findings show that interaction with ChatGPT rarely coincided with substantial changes in students’ underlying interpretations or positions. Instead, it mainly supported response expansion, argumentative reinforcement, and in some cases reformulation. A central contribution of the study is the identification of four recurrent learner profiles within the analytical category of interaction orientations: autonomy-preserving, objectivity-seeking, stance-attributing, and validation-seeking. Greater textual elaboration did not necessarily reflect stronger critical thinking or more independent interpretation. These findings highlight the need for pedagogical strategies based on source comparison, verification, and justification of generated content.