Trialling a Large Language Model (ChatGPT) in General Practice With the Applied Knowledge Test: Observational Study Demonstrating Opportunities and Limitations in Primary Care

Large language models are approaching human expert-level performance, although further development is required to match the performance of qualified primary care physicians in the AKT. Validated high-performance models may serve as assistants or autonomous clinical tools to ameliorate the general practice workforce crisis.