Breaking the Black Box of Generative AI Epistemic Justice: An Ethnography of Faculty-Student Negotiation over the Right to Explanation

The rapid integration of generative artificial intelligence (AI) into education has raised critical concerns about cognitive justice, particularly due to the “black box” nature of AI systems that limits teachers’ and students’ interpretative power and may exacerbate educational inequality. This study examines how teacher–student negotiation mechanisms can mitigate such cognitive injustice. Building on an interdisciplinary framework, we propose the concept of negotiated cognitive justice and employ a mixed-methods design informed by social cognitive theory and the technology acceptance model. Empirical data were collected from 600 teachers and students across three universities of different institutional levels in Central China, using questionnaires, behavioral logs, and controlled experiments. The findings show that democratic negotiation significantly enhances trust in AI compared to teacher-led approaches. Moreover, disciplinary background, AI usage experience, and institutional level positively moderate this effect. These findings suggest that structured negotiation can serve as an effective mechanism to restore interpretative agency and promote a more balanced integration of AI in education.