Drs spend 4x more time on paperwork than patients | UK | News

Health editor

A new study has published a sharp warning about the use of AI Chatbots to give medical advice, and revealed that ordinary people using “Doctor Boots” still cannot detect serious illnesses or make the right decisions – although the bots are at the top of medical examinations.

The news comes from the government’s ten -year plan as a part of the AI ​​has been used in health services more.

In the world’s first attempt, 1,298 people were asked to overcome ten common health scenarios such as chest pain or abdominal problems using traditional resources such as AI Chatbot (such as Chatgpt’s GPT -4O) or traditional resources such as Google or NHS website.

The boats were performed in an impressive way when tested alone – defined medical conditions correctly. But when real people used them to help, the results were worryingly poor. Only 34.5 percent of diagnostic cases were true and often – 56 percent of cases – what to do (such as going to A & E or staying at home) was taken.

Failure says how researchers from large academic institutions interact with people, not with the medical knowledge of boots. While users often provide missing or uncertain information, the boots were technically accurate, but they could not clearly explain what to do later.

The results are serious concerns for government and technology firms that encourage AI in health services.

“This is just a study, but reminds me that technology developers have rarely successfully transmitted to real life in the models, simulations and imagination of technology developers.

And Oxford University Evidence Based Medical Center Director Professor Carl Henghan said: “Doctors should take education for ten years to become a consultant for ten years, to play a role in the study and examine AI, X-rays and ECG in any way to take a role in the study and to develop a role in the study.

Last week, the government has published its 10-year plan to shift NHS from treating the disease-a largely dependent strategy to strengthen patients, applications and AIs.

In his manifesto, which was published this month, he said: “The plan (NHS) will bring it to the digital age and it will enable the personnel to benefit from the advantages and efficiency obtained from the new technology… The government will use the digital phone.

However, this study shows the gap between AI’s performance in the laboratory and its use in the real world. The authors warn that existing criteria are misleading and that AI tools should be tested not only on information but also when they communicate with non -independent communication.

“The fact that a chat boat can pass a doctor’s examination does not mean that it can help you when you are sick,” he said. “It’s like giving someone stethoscope and waiting for heart surgery.”

Researchers say that future AI tools should be much more proactive – instead of relying on users to ask users to ask the users to actively ask the conversation, to know that it is medically important.

Findings, previous studies showing that even educated doctors are not better in identifying patients with AI assistance. Now we know that the same applies to the general people.

Experts call for meticulous user attempts before using AI in health services, especially in case of direct patient advice. Without this, there is a real danger in which people can get into a wrong sense of security, look at a doctor or hurry unnecessarily to A & E.

A Ministry of Health and Social Care, thanks to our 10 -year health plan, we reduce bureaucracy throughout the health care, reduce burden administrative tasks, and use technology to spend time for doctors’ best what they do – care for patients. This includes the use of AI texts to end the need for clinical notes, letter drafts, and to end the need for manual data input so that clinicians can focus on treating patients and at the same time saving more general practice.