Medical AI reasoning can hurt clinical note accuracy

Original: When Reasoning Hurts: Source-Aware Evaluation of Frontier LLMs for Clinical SOAP Note Generation

Source: arxiv.org ↗