University Honors

Clinicians increasingly encounter AI-generated clinical information, from LLM-produced differential diagnoses to autonomous diagnostic imaging systems, yet no standardized curriculum teaches trainees to evaluate these outputs using evidence-based medicine (EBM) principles. Ophthalmology offers a particularly salient example: FDA-cleared autonomous AI systems for diabetic retinopathy screening are already in clinical use, algorithmic bias in retinal imaging threatens diagnostic equity in diverse populations, and teleophthalmology platforms depend on AI-assisted triage. These challenges are not unique to any single specialty. This project addresses the need for a transferable curriculum that teaches durable AI appraisal skills applicable across clinical contexts. "Prompt, But Verify" is a five-module asynchronous course grounded in inquiry learning and the Master Adaptive Learner model, developed at UC Riverside School of Medicine. Modules guide learners through AI foundations, EBM appraisal of AI-generated claims, patient-brought AI information, algorithmic bias identification in diagnostic tools, and clinical workflow integration. Each module follows a structured cycle of guided inquiry, curated multimedia, reflection, and peer discussion. A capstone project requires learners to produce an AI-Verified Clinical Evidence Package incorporating provenance logging, claim-by-claim verification, bias auditing, and patient-facing translation. Evaluation employs a 31-item pre/post survey with content validity evidence from subject-matter expert review, alongside rubric-scored artifacts aligned with Miller's Pyramid and Kirkpatrick's framework.