Academic Integrity in the Age of Generative AI: South African Lecturers Perspectives on AI-Generated Students Assignments

Abstract

The rapid adoption of generative artificial intelligence (GenAI) has transformed higher education by creating new opportunities for learning while simultaneously challenging traditional conceptions of academic integrity. Despite increasing institutional concern, limited empirical evidence exists on how lecturers perceive AI-generated assignments, particularly within the context of developing countries. This study investigates South African lecturers' perceptions of academic integrity in relation to AI-assisted student work through the lens of the Theory of Planned Behavior (TPB). A qualitative multi-case study was conducted involving twelve lecturers from two South African universities. Data were collected through semi-structured interviews and analysed using thematic analysis supported by NVivo 12. The findings reveal that lecturers recognise the educational potential of generative AI but remain concerned about declining authenticity, reduced critical thinking, challenges in detecting AI-generated content, and inconsistencies in institutional policies governing AI use. Attitudes toward AI, subjective norms, and perceived behavioral control jointly influenced lecturers' intentions to accept, regulate, or discourage AI-assisted assignments. The study further highlights the need to shift from detection-oriented approaches toward authentic assessment practices, AI literacy, institutional policy development, and staff capacity building. The findings extend the application of the Theory of planned Behavior to AI-mediated assessment and provide practical recommendations for universities seeking to balance innovation with academic integrity in higher education