Students in higher education are increasingly turning to Generative AI (GenAI) to assist with their assessments—both in legitimate and improper ways. This growing reliance on GenAI has significantly reshaped the academic integrity landscape, shifting the focus of academic integrity teams toward identifying potential misuse. A common consequence of such misuse is the presence of fabricated (hallucinated) or misrepresented references, stemming from GenAI’s limitations in generating accurate and verifiable information.
Manually verifying these references by reviewing original source materials is time-consuming and resource-intensive, often resulting in investigation delays and a rise in student complaints.
This poster introduces two innovative methodologies that leverage GenAI itself to detect reference-related misuse more efficiently. Specifically, I will present tailored prompt engineering techniques designed to identify both hallucinated and misrepresented citations. These approaches provide a practical framework for streamlining academic integrity investigations by harnessing GenAI’s advanced text analysis capabilities.