Not Too Harsh, Too Emotional, and Too Political: A Case Study of an ESL Graduate Teaching Assistant Enacting Agency with GenAI

Abstract This study examines how an English as a Second Language (ESL) graduate teaching assistant (GTA) named Zahra enacted her agency through the critical use of generative artificial intelligence (GenAI) while teaching freshman composition at a U.S. university. Drawing on an ecological view of agency as an achievement shaped by temporal and contextual conditions, the study analyzes three teaching-related critical incidents in which GenAI use played a significant role. Data were collected from multiple interviews with Zahra over the course of her first year. The findings suggest that although Zahra had prior language teaching experiences in her home country, she faced new demands related to unfamiliar academic genres, shifting expectations for teacher–student interaction, and the ideological sensitivities surrounding controversial topics in her instructional context. Across these incidents, Zahra used GenAI for various purposes, including consulting AI-generated examples when learning and teaching a new genre, revising the tone of an email to align with institutional norms of care, and reframing discussion questions to minimize perceived bias. These findings add nuances to the prevalent assumptions that the inclusion of AI-generated text necessarily diminishes writer agency and voice. Instead, the participant’s deliberate use of GenAI allowed her to navigate institutional constraints, establish her professional ethos, and manage the tensions emerging from her positionality as a multilingual, international GTA teaching a largely homogeneous student population. Implications call for further research on how ESL instructors mobilize GenAI to navigate evolving pedagogical, linguistic, and ideological landscapes.