Augmenting Physics Instruction and Assessment through Multimodal AI: Insights from Project Ethel

Abstract

Multimodal large language models are quickly expanding what AI can support in STEM instruction and assessment, but they also undermine answer-focused, unsupervised assessments. We describe Project Ethel, an institutional platform for course-specific, instructor-controlled AI assistance that emphasizes controlled context injection and human oversight. The paper summarizes early deployments of retrieval-augmented course chatbots, accessibility support for annotated materials, formative feedback on handwritten and computational work, and AI-assisted workflows for summative grading. For handwritten exams, we outline rubric-guided scoring and confidence-based triage (including item-response approaches) to route uncertain cases to human graders. We discuss operational constraints and design choices intended to support equitable access and course alignment while avoiding lock-in to provider-specific frontends.