GenAI Detection Tools, Adversarial Techniques and Implications for Inclusivity in Higher Education

• DOI: 10.48550/arxiv.2403.19148
• Publication Date: 2024-03-28
• Authors: Mike Perkins; Jasper Roe; Binh Vu; Darius Postma; Don Hickerson; James McGaughran; Huy Q. Khuat
• Keywords: FOS: Computer and information sciences; Computer Science - Computers and Society; Artificial Intelligence (cs.AI); Computer Science - Artificial Intelligence; Computers and Society (cs.CY)

This study investigates the efficacy of six major Generative AI (GenAI) text detectors when confronted with machine-generated content that has been modified using techniques designed to evade detection by these tools (n=805). The results demonstrate detectors' already low accuracy rates (39.5%) show reductions in (17.4%) faced manipulated content, some proving more effective than others evading detection. limitations and potential for false accusations cannot currently be recommended determining whether violations academic integrity have occurred, underscoring challenges educators face maintaining inclusive fair assessment practices. However, they may a role supporting student learning used non-punitive manner. These underscore need combined approach addressing posed GenAI academia promote responsible equitable use emerging technologies. concludes current require critical any possible implementation HE highlight alternatives strategies.