A recent study has revealed that nearly three in 10 General Practitioners (GPs) in the UK are using Artificial Intelligence (AI) tools, such as ChatGPT, during patient consultations. This rapid adoption is occurring despite widespread concerns about potential mistakes, professional liability, and the medico-legal risks associated with unregulated AI use.

The research, conducted and based on a survey of 2,108 family doctors by the Royal College of GPs alongside focus group interviews, depicts a “wild west” environment created by the absence of national regulation for AI in healthcare. This regulatory gap leaves many GPs unsure about which AI tools are safe or appropriate for clinical practice. Of the surveyed doctors, 598 GPs (28%) confirmed using AI, with usage notably higher among male GPs (33%) than female GPs (25%), and more common in wealthier areas.

GPs primarily use AI to summarize patient appointments, assist with diagnosis, and manage routine administrative tasks. Yet, a large majority of doctors, regardless of whether they personally use AI, express deep concerns about the risks of clinical errors, patient privacy breaches, and data security vulnerabilities.
Dr. Becks Fisher, a GP and the Nuffield Trust’s director of research and policy, emphasized the stark gap between government ambitions for AI to transform the NHS and the messy, uncoordinated reality of its adoption in general practice.

The fragmented regulatory landscape is made worse by inconsistent AI policies among NHS regional integrated care boards, with some approving the use of AI and others banning it entirely. Despite government expectations that AI-driven efficiencies would free GPs to see more patients, the study found that doctors instead use the time saved for self-care, rest, and reducing overtime, largely to protect themselves from burnout.

A separate study published last month in the journal Digital Health reported similar findings, noting an increase in AI use among family doctors from 20% to 25% over the past year. Dr. Charlotte Blease of Uppsala University, the study’s lead author, said that generative AI has rapidly shifted from a taboo topic to a routine tool in British medicine, highlighting the urgent need for proper training and oversight. She warned that the greatest risk lies not in GPs using AI, but in using it without clear guidance.

Patients, too, are increasingly turning to AI for health information, especially when struggling to secure GP appointments. Healthwatch England reported that around one in ten patients (9%) use AI tools for medical advice. However, the group cautioned that the quality of information varies widely, citing an example where an AI tool mistakenly confused shingles with Lyme disease.

To address these growing concerns, the government launched a national commission in September to investigate the safe, effective, and regulated use of AI in healthcare, with its recommendations expected in an upcoming report.