Early Prediction of AI-Assisted Cheating Risk in Online Exams Through Learning Analytics

Research from a study involving 52 first-year undergraduates in Turkey indicates that AI-assisted cheating risk in online examinations can be predicted early in the semester. By analyzing students' digital traces within a Learning Management System (LMS) during the initial eight weeks, a significant proportion (44.2%) of students were identified as high-risk for AI-assisted cheating, based on behaviors like copying, focus-loss, and right-click events during proctored online exams.

The ability to predict AI-assisted cheating early in an academic period offers a proactive mechanism to safeguard the integrity of online assessments and the value of credentials. This capability can inform the development of robust educational policies and technological interventions to mitigate risks, ensuring equitable and fair evaluation processes in an increasingly digital learning environment.

AI-assisted cheating presents a significant threat to the integrity of online examinations.

Forward consideration, not a verified fact.