AI-Enabled Classroom Observation: Do Data Visualizations Help Teachers Identify Attention in Classroom Video Review?

McReynolds, Andrew

Etd

AI-Enabled Classroom Observation: Do Data Visualizations Help Teachers Identify Attention in Classroom Video Review?

Public Deposited

Classroom observations are widely used to provide pedagogical feedback on instructors' classroom management techniques. However, several limitations, including observer bias and limited time for feedback review can hinder their effectiveness in helping teachers improve their instructional skills. Technological advancements have led to an increasing number of educators recording their classroom sessions for self-assessment, but classroom video review remains a labor-intensive process. While video-recorded observations address some shortcomings, challenges persist. Recent developments in neural networks and artificial intelligence offer the potential to significantly streamline the analysis of classroom dynamics through the use of eye-gaze and emotional state detectors. However, it is crucial to understand whether the information provided by these advancements is able to assist teachers in noticing nuanced classroom interactions or merely constitutes noise. This mixed-methods study examines the impact of a traditional classroom video observation compared to an AI-enhanced teacher dashboard on a viewer's ability to discern the degree of attention an instructor allocates to their students. The results show there is no difference in how reliably a participant can report the amount of attention a student received from their teacher. Interestingly, how a participant progresses through the interface is predictive of how many interactions they will have with the interface visualizations. This study contributes critical insight for developers of teacher-facing interfaces on how to improve interaction with an interface.

Creator