2024 ASEE Annual Conference & Exposition

This paper presents an innovative approach to revolutionizing STEM education by seamlessly integrating artificial intelligence (AI) into the assessment of experiment-centric pedagogy. Our research spans diverse disciplines, including biology, chemistry, physics, civil engineering, transportation engineering, mathematics, and computer science. We've transitioned from traditional teaching methods to an immersive approach, embedding experiments into core curriculum modules to convey essential concepts effectively.
Initially, this study employed the Laboratory Observation Protocol for Undergraduate STEM (LOPUS) and later transitioned to the Classroom Observation Protocol for Undergraduate STEM (COPUS), relying on manual observations. Dedicated spaces on sheets were marked at two-minute intervals to record student and instructor activities.
This study proposes a transformative leap forward, introducing an AI-based model to automate the observation process. Our primary goal is to develop a sophisticated deep learning model capable of autonomously tracking and documenting a wide range of activities performed by students and instructors in the classroom. This model will recognize, and document 26 distinct activity constructs evenly distributed between students and instructors, encompassing student questioning frequency, instructor lecturing intervals, and student-led discussions.
Leveraging state-of-the-art AI technologies, we aim to enhance the efficiency, precision, and scalability of pedagogical assessment, providing educators with invaluable insights into the dynamics of the learning environment. Our research extends beyond assessment to measure student engagement within experiment-centric classes, including the frequency of student questions, their predictive abilities concerning experimental outcomes, and participation in discussions.
In conclusion, our research drives a transformative shift in STEM education, offering a novel framework for precise assessment, personalization, and instructional enhancement. This advancement empowers educators to refine teaching strategies, enhancing student engagement, and creating a dynamic and immersive learning environment. Furthermore, the AI-based model complements existing observation protocols, like COPUS, potentially serving as a valuable control measure for assessing classroom activities.