Beyond the checklist: mapping the relational dynamics of AI ethics in higher education

Abstract Research on AI ethics in higher education has identified a wide range of concerns, including fairness, transparency, academic integrity, data security, responsibility, and dependence. Yet these concerns are often presented as separate principles, risks, or themes, leaving it less clear how they relate to each other, and how these relationships are configured by technological systems, institutional arrangements, and everyday educational practice. Drawing on an empirical study of AI ethics in Chinese higher education, this article develops an analytical mapping approach for examining these relationships. The development of the mapping approach builds on a previous analysis of semi-structured interviews with 22 participants, including 13 students and 9 teachers, from which nine ethical themes were identified: fairness, data security, academic integrity, legal awareness, interpretability, emotional care, technological dependence, critical thinking about AI, and data accuracy. Rather than treating these ethical issues as a fixed taxonomy, the mapping analysis examines them through two distinct dimensions. The first explores how each ethical issue relates to technological infrastructure, policy and practice. The second compares how strongly established technical, institutional, legal, and resource conditions shape situated action and how far situated judgement and adjustment can still materially affect outcomes. The analysis shows that ethical concerns can span multiple levels and that their location does not necessarily align with the practical capacity for action of those expected to respond. This makes visible mismatches between responsibility and control. Through this analytical mapping, the article offers an empirically grounded, context-sensitive approach for examining how AI-related ethical concerns are located, constrained, connected, and redistributed across higher education settings.