The study highlights the potential for students to cheat using AI undetected
A University of Reading study has found that AI-generated exam answers from fictitious students outperformed real student submissions, often going undetected by markers.
Researchers used ChatGPT to create answers for 33 fake students in an undergraduate psychology course. AI answers averaged half a grade higher than those of real students, with 94% not raising suspicion among markers.
The study, published in Plos One, highlights the potential for students to cheat using AI undetected.
Researchers, including Associate Prof Peter Scarfe and Prof Etienne Roesch, call for global education sectors to adapt in response to AI's impact on assessment integrity.
Read full article