The Polemics of Ethical AI Usage in Higher Education: A Case Study Investigating the Axiological Expectations Mismatch

Abstract

The rapid integration of generative artificial intelligence (GenAI) into higher education has created significant tensions around authorship, assessment authenticity, and academic integrity. This paper introduces axiological expectations mismatch as a conceptual framework for understanding a core governance problem: the divergence between institutional value frameworks and the ethical reasoning students apply when using AI in their academic work. Drawing on an interpretive qualitative case study at a single private higher education institution in South Africa, the study analyses twelve institutional documents produced between 2021 and 2025, supplemented by descriptive trend data from 3,854 plagiarism incidents over the same period. Schwartz’s (2012) theory of basic values provides the theoretical lens; reflexive thematic analysis is the analytic method. Three findings emerge: first, a shift from prohibition to conditional permission for AI use, contingent on disclosure and authorship accountability; second, a reframing of academic integrity as a developmental process rather than a purely disciplinary matter; and third, evidence that policy adaptation improved institutional capacity to recognise and classify AI-related misconduct before it reduced its incidence. The paper argues that AI-related integrity disputes are better understood as conflicts between competing values (fairness, accountability, efficiency, and innovation) than as individual moral failings. Implications for policy design, assessment reform, and faculty development are discussed.