Learning Progression-Guided AI Evaluation to Support Feedback on Multimodal Scientific Modeling Tasks

• In Event: AI facilitated feedback to support learning

Abstract

Providing cognitively appropriate feedback to individual learners is key to supporting science understanding. The process of doing science is multimodal, and we should support understanding via multiple modalities (e.g., written explanations, drawings). Developing feedback for supporting multimodal proficiency is challenging. Machine learning has been utilized to score and provide feedback on open-ended models and explanations. However, the challenge of aligning multiple modalities for fostering understanding in a developmentally-appropriate way remains. We advance prior work on AI evaluation of stand-alone models and explanations and leverage a multimodal learning progression (LP) for training AI to analyze assessments for delivering personalized LP-aligned feedback on multimodal tasks. We test the feedback with students to gauge how it helps them revise their responses.

Authors

• Leonora Kaldaras, Texas Tech University

• Clare Carlson, Michigan State University

• Wenxiu Tang, Michigan State University

• Tingting Li, Washington State University

• Prudence Djagba, Michigan State University

• Joseph S. Krajcik, Michigan State University

• Kevin Haudek, Michigan State University