Angeliki Pantazi

🚨 I just read the most important AI paper of 2025 and it might redefine how medical students learn. A new paper from Dartmouth and Stanford shows that generative AI can finally deliver personalized tutoring at scale without hallucinations. They built a system called NeuroBot TA, an AI teaching assistant powered by Retrieval-Augmented Generation (RAG). It only answers from instructor-approved materials, meaning it’s accurate and contextual. The results? • 329% spike in student usage before exams • Heavy after-hours use (post–5 pm) • Most questions on neuroanatomy & clinical disorders • Students trusted it more because it cited real sources but some wanted it less constrained The tension is clear: Reliability vs. breadth. Students love AI that’s grounded in truth, but they still crave flexibility. This is what the future of med ed looks like: precision learning tailored, verified, 24/7 tutoring built directly from your course. Full paper: npj Digital Medicine (Nature, 2025)