By D. Maan, Jadetimes News
AI Outperforms Real Students in University Exams, Raising Concerns About Academic Integrity
In a limited study conducted by University of Reading researchers, artificial intelligence outperformed real students in university exams. Researchers created 33 fictitious students and used the AI tool ChatGPT to generate answers for undergraduate psychology exams. The AI generated responses achieved grades half a boundary higher on average than those of actual students.
Furthermore, 94% of the AI generated essays went undetected by markers, with only a 6% detection rate, which the study suggests might be an overestimate. The findings, published in the journal PLOS One, highlight the potential for students to use AI to cheat and achieve higher grades without being caught.
Associate Prof. Peter Scarfe and Prof. Etienne Roesch, who led the study, emphasized the significance of these results for educators worldwide. Dr. Scarfe pointed out that many educational institutions have shifted away from traditional exams to promote inclusivity in assessments. However, the study underscores the necessity for the global education sector to adapt in response to AI's impact on the integrity of educational assessments. While a complete return to handwritten exams is unlikely, evolving assessment methods will be crucial to maintaining academic standards.
AI Performance in University Exams and the Challenge of Abstract Reasoning
In a study, researchers submitted fake exam answers and essays for first, second and third year modules without the marker's knowledge. The AI generated responses outperformed real undergraduates in the first two years. However, third year students scored higher than the AI, aligning with the researcher's observation that current AI struggles with more abstract reasoning. This study is the largest and most robust blind study of its kind to date.
Academics are increasingly concerned about AI's role in education. For instance, Glasgow University recently reinstated in person exams for one course to counteract AI influence. Additionally, a study reported by the Guardian earlier this year found that most undergraduates used AI programs to assist with their essays, but only 5% admitted to directly pasting unedited AI generated text into their assessments.
