Can Artificial Intelligence Complete My Assessment? A Student Led Initiative to Stress Test the Academic Integrity of University Assessment Using Generative AI

Abstract

[EN] The ability of Generative AI (GenAI) to perform complex tasks has caused mixed feelings within the field of education. The most significant concern is the implications of GenAI for academic integrity. In this study, students applied GenAI to complete past university assessments adapted as research tests, with the goal of achieving a pass grade when graded by an academic, undetected by an AI writing detection tool. The study reveals that from a sample of 21 valid research tests, 23.8% (5) passed when graded by an academic achieving grades between 40% and 60%, with AI writing detection scores ranging between 0% and 14%. AI writing detection scores lower than 20% are potentially a false positive in the context of investigating breaches of academic integrity. The researcher concludes that many traditional methods of assessment in universities are obsolete in the face of increasingly undetectable AI generated solutions.