AI and Exam Integrity in Cameroon’s Bilingual GCE: A Qualitative Conceptual–Analytical Study

Abstract

The rapid spread of artificial intelligence (AI) tools, including text generators and translation applications, presents new challenges to the authenticity of high-stakes examinations. Cameroon’s bilingual General Certificate of Education (GCE) is particularly exposed due to its dual-language structure and growing student access to AI-assisted writing. This qualitative, conceptual–analytical study examined risks of AI-assisted malpractice using three data sources: official GCE policy documents, written reflections from ten experienced secondary-school teachers and examiners, and an analysis of selected A-Level French examination papers from 2019 to 2025. Data were analysed thematically with a focus on patterns of AI susceptibility, linguistic irregularities, and vulnerabilities in question design. Findings indicate three main risk patterns: (1) automated translation producing near-perfect responses inconsistent with candidates’ general language performance; (2) AI-generated essays enabling structurally coherent but cognitively shallow scripts; and (3) exam items relying on general knowledge showing high susceptibility to machine-generated answers. Analysis of 224 scripts revealed several cases where translation scores exceeded 90% while essays in the same paper fell below 10%, suggesting potential AI involvement. Existing regulations did not explicitly address AI-assisted authorship, and invigilation systems showed limited capacity to detect covert digital tools. AI tools create identifiable authenticity gaps in Cameroon’s bilingual GCE by enabling discrepancies between candidates’ productive skills and their written outputs. Current safeguards insufficiently address these risks, indicating a need for updated policies, examiner training, and improved detection capacity to protect examination validity.