GP or ChatGPT? Ability of large language models (LLMs) to support general practitioners when prescribing antibiotics

Large language models (LLMs) are increasingly used for diagnosis and treatment of antibiotics. A study involving 24 vignettes from four countries showed high accuracy in GPs’ diagnosis and prescribing of antibiotics. However, GPs often failed to reference and prescribe according to national guidelines. LLMs scored high in diagnosis, antibiotic prescribing, and antibiotic choice, but often failed to reference correctly. Data leakage was also a concern. The study suggests that GPs are best placed to interpret vignettes, apply national guidelines, and prescribe the right dose and duration.