One in four Brits would trust AI over their doctor if the two gave different medical advice, according to new research.
More than half of people are now turning to AI for medical advice, the Medichecks survey revealed, with 55.7 per cent of respondents saying they had already used tools such as ChatGPT, while 95.7 per cent said they would be willing to do so again.
When asked what they would do if AI and their GP gave conflicting advice, 41.7 per cent said they would not know who to believe, while 23.8 per cent said they would follow AI’s advice over that of their doctor.
Dr Natasha Fernando, medical director at Medichecks, said: “AI can be a helpful tool for learning more about your health or preparing questions before a medical appointment, but it should never replace proper medical care.
“A GP doesn’t just look at symptoms in isolation, they consider your full medical history, examine you in person, assess risk and make decisions based on clinical training and accountability.”
“Medicine is rarely black and white. Two people can present with very similar symptoms but require completely different care.
“AI cannot interpret the full context of someone’s health, their lifestyle, their history, or the subtle warning signs a clinician might pick up during a consultation.”
Nearly three-quarters of people aged 25 to 34 said they had used AI for a medical query, compared with 34.9 per cent of people aged over 75.
Men were also more likely than women to rely on AI for health advice. In the survey, 63.7 per cent of men said they had used it for a medical concern, compared with 50.9 per cent of women.
Women were also more likely to express concerns about healthcare becoming less personal if AI tools become more widely used.
When looking to the future, more than half of respondents, 51.7 per cent, said they believe AI will be better than doctors at making accurate health predictions by 2030.
Only 19.1 per cent said they would trust human doctors more.
Research suggests today’s generative AI systems can already achieve diagnostic accuracy levels similar to non-expert doctors, averaging around 52 per cent, while some specialist AI diagnostic tools have reported accuracy rates as high as 85 per cent in complex cases.
Despite the surge in public use of AI tools, the technology is already embedded in parts of modern healthcare. Across England, AI is used in stroke units to help analyse brain scans, while radiology departments use it in clinical practice.
An AI system called Mia, used to analyse mammograms at NHS Grampian, reviewed scans from more than 10,000 women and detected every case of breast cancer identified by radiologists, as well as 11 additional cases that had not initially been spotted.
However, AI systems can also make serious errors.
Google’s Med-Gemini model reportedly identified abnormalities in a non-existent body structure called the “basilar ganglia”, highlighting the risk of so-called hallucinations where AI generates incorrect information.
Fernando said: “AI does not take responsibility for the outcome of your care. Doctors do. That accountability is fundamental in medicine.
“If you have worrying symptoms, the safest course of action is always to speak to a healthcare professional who can properly assess you.
“AI may become an increasingly valuable tool within healthcare, particularly for analysing large datasets or supporting clinicians with decision-making.
“But the future of medicine will almost certainly involve humans and technology working together, not technology replacing doctors.
“AI can help people become more informed about their health, which is positive.
“But it should be used as a starting point for discussion with a healthcare professional, not the final word.”
