Detection of AI-generated Writing in Students’ Assignments: A Comparative Analysis of Some Tools' Reliability

This article answers the question: How reliable are current detection tools at identifying human or AI writing in students' assignments? It aims to test the reliability of these tools through a comparative analysis of 10 of the most popular ones. This enabled us to assess the reliability and robustness of such tools in the face of various writing manipulations that some students may perform while producing their work to hide the artificial origin. We revealed the limitations of each tool taken individually and the need to combine several to overcome their deficiencies and use them to detect the presence of AI writings in students' work.