Researchers evaluated three large language models for medical consultation and found that they often provide self-care advice before the patient's concern is clearly stated. This 'preformulation gap' can lead to misdiagnosis or inadequate treatment. The study suggests that these models should be evaluated based on their ability to elicit decisive facts from patients, rather than just their accuracy in providing final answers. --- Why it matters: This matters because it highlights a critical flaw in the current evaluation methods for medical consultation LLMs. If these models are not designed to handle vague or unclear patient concerns, they may provide suboptimal care, which can have serious consequences for patients' health. Source: https://arxiv.org/abs/2608.17330