A new study by University of Reading published in PLOS One finds that many students submit their AI generated exams and most of them go undetected too. The researchers of the university secretly submitted AI generated exams in addition to real exams done by students of psychology and made some examiners check the papers. The results found out that 94% of the AI generated submissions weren’t detected by examiners.
This wasn’t all because the AI generated answers didn’t only get passed by human examiners, but they also got more scores as compared to exams by human students. In some cases, a difference of full grade was also observed between AI generated exams and human written exams. Some AI generated exams also got first-class honors.
This study has raised serious questions about the future of education. As AI is growing fast and is used by a lot of people, how can colleges and universities administer AI detection methods? Many of the AI detection tools are inadequate and cannot detect AI writing efficiently. Many of the educational institutions are not using traditional exam procedures and are making their students write assessments. Assessments can easily be written using AI tools but no institute is prepared to detect AI in those assessments.
An associate professor at Reading’s School of Psychology says that educational institutes shouldn’t go back to hand-written exams but they should work to evolve with AI. The researchers say that this study should be seen as a call for action because if the students are using AI tools in their education, how can they work in a reputable institute without having any knowledge about the subject they studied. Educators and administrators should think seriously about it because the institution of education is not in a good place right now.
Image: DIW-Aigen
Read next: The best AI content detectors in 2024
by Arooj Ahmed via Digital Information World
No comments:
Post a Comment