IDENTIFYING OPTIMAL AI MODELS FOR ONLINE EXAMINATION PROCTORING: A SYSTEMATIC LITERATURE REVIEW

Online exam proctoring solutions have become more popular in higher education due to digital change, particularly during the COVID-19 epidemic. These methods do, however, present issues with user privacy, ethics, and detection accuracy. This study focuses on the use of artificial intelligence (AI) models in online proctoring by conducting a Systematic Literature Review (SLR) of 50 peer-reviewed articles published between 2020 and 2025. Face recognition, Convolutional Neural Networks (CNN), YOLO, Long Short-Term Memory (LSTM), eye tracking, Haar Cascade, and hybrid techniques are among the most widely used models. The results show that while YOLO and CNN exhibit remarkable efficiency and accuracy in real- time visual detection, they nonetheless raise ethical questions. Although they demand more resources, hybrid models provide a more balanced approach by combining AI skills with human oversight. The study comes to the conclusion that there is no one model that is always the best option; rather, the best option is determined by the test setting, the size of implementation, and the trade-offs between technical efficiency, monitoring accuracy, and ethical issues. These results lay the groundwork for creating AI-driven proctoring systems that are more sustainable, flexible, and egalitarian