Language models are susceptible to incorrect patient self-diagnosis in medical applications

Open in new window