An AI-Assisted Interactive Video-Based System (AI-LabEyes) for Enhancing Visualization, Formative Feedback, and Collaborative Learning in Chemistry Laboratory Instruction

Abstract

This study describes AI-LabEyes, an AI-assisted, multiangle video system designed to support observation, formative feedback, and teacher-mediated evaluation in secondary-school chemistry laboratories. The system combines synchronized user-facing, side, and overhead video views; computer-vision-based identification of selected titration actions; secure storage for deidentified instructional clips; and tools for teacher or expert review. In a field implementation during a high-school acid–base titration competition, AI-LabEyes scores from 46 valid student videos were compared with teacher-adjudicated scores for six operational actions. Overall agreement was 88%, with stronger performance for visually clear actions such as leak inspection and buret conditioning (96% accuracy for each) and weaker performance for actions requiring finer temporal or judgment-based interpretation. The system effectively promotes the visualization of experimental teaching, formative feedback of students, and collaborative improvement, providing a reference for AI-assisted experimental teaching.