Douglas Flora, Executive Medical Director of Yung Family Cancer Center at St. Elizabeth Healthcare, President-Elect of the Association of Cancer Care Centers, and Editor-in-Chief of AI in Precision Oncology, shared a LinkedIn post noting that a system capable of clearing the U.S. medical licensing exam with a ninety-first percentile score would miscount the r’s in “strawberry,” answering two instead of three. Flora explained that these tools process the world in fragments called tokens and were never built to count letters. In his essay, he discusses how retrieval-augmented generation (RAG) keeps chatbots grounded in institutional protocols, and how systems have become agentic, transitioning from answering questions to drafting notes or matching oncology patients with clinical trials. Flora argues that healthcare leaders must hold both a genuine wonder for these tools’ capabilities and a healthy doubt about their limitations, emphasizing that the most critical component of any new medical tool will always be the doctor sitting beside it.