An online survey conducted by the BMJ revealed that one in five UK GPs are incorporating generative AI tools, such as ChatGPT, into their clinical practice despite a lack of formal training.
Published on 17 September 2024, the survey disclosed that 205 out of 1,006 respondents (20%) are utilizing AI tools in their practice. Of those affirmative responders, 29% are using AI tools to document patient appointments, and 28% are employing them to suggest differential diagnoses.
The survey identified ChatGPT as the most commonly used large-language model (LLM) among GPs.
Dr. Charlotte Blease, the study’s author and an associate professor at Uppsala University in Sweden, suggests that GPs could derive value from these tools, especially for administrative tasks and clinical reasoning support.
Although beneficial, Dr. Blease acknowledges that these tools have limitations as they can introduce subtle errors and biases.
Despite the lack of training and clear work policies, GPs continue to incorporate AI tools into their practice. Dr. Blease emphasizes the importance of educating physicians and trainees on both the benefits and risks associated with these tools.
Concerns regarding patient privacy were also raised as the information gathered by AI tools may be shared with internet companies without clear consent from patients.
Research commissioned by the Health Foundation in July 2024 indicated significant support from NHS staff and the public for using AI in patient care.
While there is excitement surrounding generative AI in healthcare, Dom Cushnan, director of AI, imaging, and deployment at NHS England, stressed the importance of evidence-based and clinically appropriate use of AI tools in health systems.
