A fifth of GPs are using AI despite a lack of guidance or clear working policies, a British survey has found

A fifth of GPs appear to have readily integrated AI into their clinical practice, despite the lack of formal guidelines or clear working policies on the use of these tools, according to findings from an online UK snapshot survey, published in the open access magazine BMJ Health and Care Informatics.

Doctors and medical trainees should be fully informed about the pros and cons of AI, especially due to the inherent risks of inaccuracies (‘hallucinations’), algorithmic biases and the potential to compromise patient privacy, the researchers conclude.

Following the launch of ChatGPT in late 2022, interest in chatbots with large language models has surged and attention has increasingly focused on the clinical potential of these tools, the researchers say.

To gauge the current use of chatbots to support any aspect of clinical practice in Britain, the researchers distributed an online survey in February 2024 to a randomly selected sample of GPs registered with clinical marketing service Doctors.net .uk. The study had a predetermined sample size of 1,000.

Physicians were asked whether they had ever used any of the following in any aspect of their clinical practice: ChatGPT; Bing AI; Google’s Bard; or ‘Other’. And then they were asked what they used these tools for.

Approximately 1,006 GPs completed the survey: just over half of the responses came from men (531; 53%) and a similar percentage of respondents (544; 54%) were 46 years or older.

One in five (205; 20%) respondents reported using generative AI tools in their clinical practice. Of these, more than one in four (29%; 47) reported using these tools to generate documentation after patient appointments, and a similar percentage (28%; 45) said they used these tools to generate another diagnosis for to set. One in four (25%; 40) said they used the tools to suggest treatment options.

The researchers acknowledge that the survey respondents may not be representative of all UK GPs, and that those who responded may have been particularly interested in AI – for better or for worse – potentially introducing some degree of bias into the findings introduced.

Further research is needed to learn more about how physicians use generative AI and how best to implement these tools safely into clinical practice, they add.

“These findings indicate that primary care physicians can derive value from these tools, especially in administrative tasks and to support clinical reasoning. However, we caution that these tools have limitations as they can embed subtle errors and biases,” they say.

And they point out: “[These tools] could also risk causing harm and undermining patient privacy, as it is not clear how the internet companies behind generative AI use the information they collect.

“While these chatbots are increasingly the target of regulatory efforts, it remains unclear how legislation will intersect in a practical way with these tools in clinical practice.”

They conclude: “The medical community will need to find ways to educate both physicians and trainees about the potential benefits of these tools in summarizing information, as well as the risks in terms of hallucinations. [perception of non-existent patterns or objects]algorithmic biases and the potential to compromise patient privacy.”