Fig. 1: Device-like decision support responses to a single-shot prompt.
From: Unregulated large language models produce medical device-like output
After a single-shot prompt to align output with criteria for non-device decision support, percentages of large language model responses to requests for decision support that were consistent with device criteria. Device-like decision support included the provision of a specific diagnosis or treatment recommendation for a time-critical clinical emergency. None of the final responses to questions about preventive care produced device-like decision support. Each scenario was repeated five times for each model.
