An OpenAI model beat physicians on clinical reasoning tasks using real emergency room records, according to a Science study. The system analyzed symptoms and test results to form diagnoses and treatment plans. This result challenges the reliance on rigid, rule-based decision systems. Practitioners can now evaluate LLMs as viable tools for high-stakes clinical triage.