Researchers at the University of Reading made 33 fake students, then used ChatGPT, an AI tool, to write answers for module examinations for the university’s undergraduate psychology program.
According to them, the AI students’ average scores were half a grade point higher than those of their peers in the real world.
Furthermore, 94% of the AI writings did not raise any worries about markers, and they “verged on being undetectable.”.
The study, which was published in the journal Plos One, indicates that the 6% detection rate is probably an overestimate.
As a result, students could use AI to cheat covertly and outperform their peers in terms of grade.
The study’s lead authors, Associate Professors Peter Scarfe and Etienne Roesch, stated that their results need to serve as a “wake-up call” for educators everywhere.
Dr. Scarfe continued: “Many institutions have moved away from traditional exams to make assessment more inclusive.”Our findings demonstrate the significance of comprehending how AI will impact the integrity of educational assessments on a global scale.
