Addressing the ChatGPT Crisis: The Rise of AI-Generated Academic Assignments in Higher Education

Description

Sonayena Kenneth Nokele (Department of Public Administration & Management, College of Economic and Management Sciences University of South Africa, South Africa)

This paper examines the “ChatGPT crisis” as a higher education pedagogy, policy, and justice issue. Based on a structured review of peer-reviewed scholarship (2020-2025) in education, assessment, and technology studies, it synthesises evidence on three domains: AI-generated academic work, institutional challenges in detection, equity, and workload, and sectoral responses by universities, policy makers, and technology developers. Large language models separate textual fluency from competency, reducing the evidence value of take-home essays and short-answer activities. Multilingual students face due-process concerns from detection-centric replies, since classifier performance is variable between prompts and disciplines and mistake rates, notably false positives, are high. Case management, assessment redesign, and staff development expenditures are growing for institutions. Comparative policy mapping shows a change from prohibition to conditional authorisation based on transparency, attribution, and person comprehension. The paper proposes a design-for-integrity approach that moves assurance from post-hoc policing to provenance-rich assessment using Academic Integrity Theory, Technological Determinism, and AI-ethics fairness. Strategic recommendations prioritise process-evidenced task sequences (drafts, artefacts, micro-vivas), explicit AI-use statements in briefs and rubrics, scaled AI-literacy for staff and students, and a governance stance that triangulates detector outputs with process and oral evidence. In order to maintain academic integrity, pedagogy, governance, and technology must be integrated to make authorship visible, normalise bounded augmentation, and safeguard fairness while maintaining strict standards.