Creating an artificial intelligence model to improve student engagement based on teacher’s behaviours and movements in video conferencing

Creating an artificial intelligence model to improve student engagement based on teacher’s behaviours and movements in video conferencing

The prevalence of online learning in higher education is driven by its numerous benefits, such as accessibility, flexibility, cost efficiency, and the global impact of COVID-19. Despite these advantages, lack of student engagement arises as a significant challenge. This study aims to address this issue by providing an Artificial Intelligence (AI) model designed to improve online student engagement based on teachers' behaviours and movements in video conferencing. A design-based research (DBR) approach was employed to develop an AI model capable of auto-generating reports on engagement-enhancing teachers' behaviours and movements as characteristics and indicators of engaging teaching videos. The study is structured with three phases. In the initial phase, a systematic literature review identified 11 characteristics and 47 indicators of engaging teaching videos. During the second phase, an AI model was developed and trained using the identified characteristics and indicators under the guidance of an AI expert. Prototype 1 (model 1) was trained using manual annotations of 25 Zoom-recorded lectures. Prototype 2 (model 2) was refined using oversampling techniques to address data imbalance and misleading metrics obtained in Prototype 1 (model 1), resulting in improved model performance. In the final phase, the AI model underwent three levels of evaluation, analysing and comparing its reports with those generated by human experts. Comparison results revealed a low similarity rate between the AI model and the expert reports, indicating the need for further refinement to align the model's performance more closely with human experts. This research contributes a comprehensive list of characteristics and indicators of engaging teaching videos, aiding in enhancing online student engagement. The provided manual annotation procedure supports similar AI model development in the future. Moreover, collaboration between educators (experts) and AI engineers to develop and evaluate AI models in education is emphasised, thereby breaking the barriers between educators and AI scientists. The reports generated by the model will also support educational institutions in promoting continuous enhancements in teaching methods and practices.

File reproduced in accordance with the copyright policy of the publisher/author/creator.