Loading...
Readily available, fluent, effective, and personalized AI is now assisting a majority of clinicians in the U.S. But all large language models (LLMs) make mistakes, including clinician decision-support tools that cite research articles and guidelines. So how should clinicians use AI at the bedside effectively and safely? Three hospitalists with expertise in clinical decision-support principles offer a practical, human-centered framework: the 5Rs.
Right task: Know when to use an LLM and when not to. Use an LLM to broaden your differential; refrain from using it to analyze an ECG (today’s LLMs have limited processing abilities).
Right model: Match the tool to the task. Use “fast thinking” models (e.g., ChatGPT Instant) for trans…