Exploring Academic Integrity in the Age of Generative AI

Anthony Ward, Sathiamoorthy Manoharan, Xinfeng Ye

Abstract: Traditionally, statistics suggest that around 30% of students resort to cheating when presented with the opportunity. This alarming trend often goes unnoticed by instructors, as many do not actively monitor for such misconduct. However, the onset of the pandemic saw a surge in cheating incidents, mainly due to the shift to online learning environments where supervision was more challenging. Despite this, instances of cheating remained detectable, albeit with some effort. As we transition out of the pandemic era, a new challenge has emerged in the form of generative AI technology, which has dramatically altered the landscape of academic dishonesty. With the advent of sophisticated generative AI models, it has become virtually impossible to discern between authentic student work and AI-generated content in unsupervised settings. Consequently, there has been a noticeable decline in detected cheating cases compared to pre-pandemic levels, largely attributed to the widespread use of generative AI tools. Traditional plagiarism detection software such as Turnitin and MOSS, once considered reliable safeguards against academic dishonesty, have become obsolete in the face of these advanced AI systems. The unique solutions produced by generative AI render such detection methods ineffective, posing a significant challenge to maintaining academic integrity in the modern educational landscape. This paper explores the nature of academic misconduct incidents in the context of two universities across the globe: one in the United Kingdom and the other in New Zealand.