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

Major Thesis Advisor

Associate Professor Jacob Whitehill

Thesis Reader

Assistant Professor Stacy Shaw

Abstract

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.

Note: This material is based upon work supported by the National Science Foundation under Grant No. IIS-1822768. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.

PI: Jacob Whitehill, PhD

This work was supported by the National Science Foundation under Grant No. IIS-1822768

For the 2022-2023 academic year, I am co-chairing the WPI Learning Sciences and Technologies Colloquium Series with Ph.D. Student Kirk Vanacore. 

This year, the goal for the colloquium series is to provide students and faculty with:

1. Presentations and discussions with different researchers in academia & industry; 

2. Access to workshops to expose our community to important topics such as Open Science, Statistical Modeling, and more;

3. Presentations by graduate students regarding their recent progress on their Theses and Dissertations; and

4. Monthly social hours where students can meet each other and build connections across our multiple labs.