Real-Time AI Integrity: Enhancing Cheating Detection through Neural Network-CNN Data Analysis

The rising demand for online learning and the high academic reliance on artificial intelligence (AI) have led to the need for the development of an automated system that detects and mitigates instances of student cheating, especially in large settings, to protect the academic process. The dramatic shift towards online learning, especially after COVID-19, has urged the development of such systems to ensure exam integrity. This research aims to develop an advanced Neural Network-CNN framework to detect and predict cheating in online and large-scale lab assessments. The proposed framework aimed to detect cheating through specific measures and patterns depending on students’ actions or responses, such as eye blinking, attention, lip movement, and time-automated data series. This research employed various data inputs through a complex neural network architecture during assessments, and the framework uses a multi-modal approach that analyzes all the interactions of the students and their patterns. The results show that the AI framework reduces up to 41.2% of cheating among n students and can predict cheating up to 35.3% among 100 students in one exam.