Doctors say they’ve been left to confront a “wild west” of AI, with no guidance on which tools to adopt or how to use them, despite the technology becoming increasingly critical for GPs to manage soaring workloads.

New research from the Nuffield Trust found that more than a quarter (28%) of UK family doctors are now using AI tools in their daily work over a range of tasks, including producing clinical notes and referral letters, triaging patients based on clinical need, and helping trainees.

Polling more than 2,100 GPs alongside the Royal College of General Practitioners (RCGP), the Nuffield Trust found that while 13% of doctors have been provided AI tools by their practice, 11% have been forced to obtain them independently, resulting in a reliance on public tools such as ChatGPT.

This, said the Nuffield Trust, not only raises serious concerns about patient safety, but could also worsen health inequalities, as practices in deprived areas are less likely to provide doctors access to relevant AI tools.

Of the 1046 GPs who said they worked in more deprived areas, the study found only around a quarter (27%) used AI tools compared to more than a third (35%) of the 467 GPs working in more affluent areas.

The Nuffield Trust argues that GP use of AI currently depends on a “postcode lottery” of local policies, primary care networks and Integrated Care Boards, with some ICBs forbidding all AI use, while others actively encourage pilot programmes.

As a result of this patchwork approach, GPs are often left to make decisions about which tools to use based on peer experience or direct marketing from AI companies, with doctors becoming more worried about patient safety, professional liability, data privacy and patient consent.

However, with general practice under immense pressure, and the burden only increasing, AI is becoming a vital resource in helping GPs reduce overtime and admin, while benefiting doctor-patient relationships. For example, AI scribes allow GPs to engage more meaningfully during appointments while still producing adequate consultation records. 

But according to the report, regulation and training aren’t happening fast enough, with the pace at which the tech is being adopted clashing against sluggish NHS bureaucracy.

The Nuffield Trust said that despite the government’s high hopes that AI can make the NHS more productive, its 10 Year Health Plan lacks clear targets and has a limited roadmap for how to get there, with a mismatch between the perceived benefits for GPs and policy ambitions, largely focused on boosting productivity.

Recommended reading

• NHS to Trial AI at ‘Unprecedented Scale’
• Data Silos Preventing Healthcare Orgs from AI Gains
• Will AI in Healthcare Leave Disabled Britons Behind?

Calling for clearer direction for GPs, the Nuffield Trust said it was urging policymakers and tech firms to come together with doctors to establish consistent national guidance, improving training, and support best practice to improve working conditions and patient outcomes.

“AI clearly has the potential to dramatically change the way healthcare is delivered for the better, but using AI in practice is certainly not without risk, with patient safety and data protection at stake if robust safeguards and regulation are not in place,” said Professor Victoria Tzortziou-Brown, chair of the Royal College of GPs.

“If AI is going to help GPs deliver high-quality care for patients, it is paramount that there is further investment into modern IT systems, consistent national policy and regulations, and the right support so that practices can adopt new technologies safely and effectively.”