Large language models in clinical practice

Large language models (LLMs) are reshaping mental health care from both sides of the clinical encounter. Patients are increasingly turning to large language models (LLMs) for mental health support, often before, alongside, or instead of professional care. At the same time, health systems are actively exploring how LLMs can be safely and effectively integrated into clinical workflows, from documentation and decision support to patient communication and clinical reasoning. This raises urgent questions for psychiatric practice: What does the widespread use of LLMs for mental health support demand of a profession still developing the frameworks to evaluate them and what needs to be done to integrate LLMs safely into clinical workflows of clinicians?

In this session Dr. Herpertz examines the evidence on how widely LLMs are used for mental health purposes in the general population and among people actively seeking mental health support, and what risks this may carry. He then presents data on how lay people interact with LLMs when confronted with mental health questions, and whether conversations with chatbots help them arrive at appropriate understanding and courses of action. Finally Dr. Herpertz turns to the clinical workforce and outlines what clinicians need to be taught in order to integrate artificial intelligence effectively and safely into their workflows.