Unpacking how humanities and social sciences undergraduate students maintain agency in GenAI-assisted learning

The integration of Generative AI (GenAI) into higher education raises critical questions about how students maintain intellectual agency while engaging with generative systems, particularly in the humanities and social sciences (HSS), where interpretation, argumentation, and authorship are central. Addressing this issue, this study employs a grounded theory approach to examine how HSS undergraduate students sustain agency in GenAI-assisted learning. Semi-structured interviews were conducted with 37 students across multiple disciplines, and data were analyzed using open, axial, and selective coding. Based on this analysis, the study develops the Strategic Engagement and Critical Evaluation model, which conceptualizes agency as an ongoing regulatory process enacted through critical assessment of AI outputs, boundary setting, and adaptive prompting. The findings indicate that students operate as “humans in the loop,” continuously calibrating trust while retaining evaluative authority over AI-generated content. By offering a process-oriented explanation of agency in human-AI interaction, the study also identifies implications for explainable AI design and educational policy.