Navigating Health Information in the Age of AI

From Googling symptoms to asking ChatGPT for medical advice, the way we look for health information is changing rapidly.
Here, GP Dr Vicky Carre explores the risks and opportunities of AI in healthcare, including how to recognise misleading information, how to use AI safely, and how we’re already using the technology at Healthcare Group to improve patient care.

I went along to the Guernsey Chamber of Commerce Gala Dinner last week, celebrating a series of events exploring the opportunities that artificial intelligence (AI) could bring to Guernsey businesses and the wider community.
The speed of AI development is both frightening and exciting in equal measure. It is easy to feel threatened by the technology, particularly when we are bombarded with scare stories about AI bots taking over our jobs or even the world. However, perhaps a more constructive approach is to consider how we can embrace this technology and use it to enhance our lives, complementing rather than replacing what we do. The potential efficiency savings for businesses are enormous, and maybe we could all continue to perform our adapted jobs but enjoy a three-day weekend!
Healthcare is one area where the potential is enormous. AI is already changing how healthcare professionals work, how diseases are diagnosed and how treatments are developed. It is also changing how patients look for information and make decisions about their health.
From Dr Google to Dr ChatGPT
The way we access health information has changed dramatically over the past couple of decades. Not so long ago, if we had a health concern, our first action was to get a GP appointment. Then came Google, and suddenly information about any symptom, condition or treatment was at our fingertips. We moved from patients bringing in clippings from the Daily Mail health column to bringing in reams of printouts from the internet.
There were real benefits. Patients became more informed and engaged, could research treatment options and arrive at appointments with questions. Online forums and support groups also gave people living with particular conditions the opportunity to form communities and share experiences.
But there was a downside. The internet does not reliably distinguish between good evidence and somebody's opinion, and online forums can spread inaccurate advice. Some websites also have commercial interests, which are not always obvious. And a search for a headache can quickly lead to a list of alarming possibilities.
Social media then took things a step further. Rather than actively searching for information, health advice began appearing in our feeds, often delivered by influencers, celebrities or people presenting themselves as experts. TikTok and Instagram have become particularly influential, especially among younger people.
There is some excellent health education online, but there is also a great deal of misinformation. A confident delivery, a large following or a celebrity endorsement does not make something medically accurate. Sponsored content and commercial interests can also blur the distinction between education and advertising. Even a recognisable medical face is no guarantee of reliable information. Just last week, it was reported that television doctor Dr Rangan Chatterjee and his company attracted an Advertising Standards Authority ruling over advertising for a blood-testing programme. His claims that blood tests offered by his own company could identify markers of Alzheimer's disease were not supported by sufficient evidence.
Even Google itself is changing. Many people may not realise that the AI-generated summaries now appearing at the top of Google search results, known as AI Overviews, are produced using generative AI. You can even move seamlessly from one of these summaries into a conversational AI experience, asking follow-up questions without necessarily appreciating that you have moved beyond a traditional Google search. The distinction between searching the internet and interacting with AI is becoming increasingly blurred.
And now we have AI chatbots. Instead of just reading information, AI gives us something that feels like a personal medical conversation. We can ask ChatGPT a question, describe our symptoms, upload a test result and ask follow-up questions. The response can adapt to what we tell it, explain unfamiliar medical terminology and present information in a way that feels tailored to us.
Used sensibly, this can be genuinely helpful. It might help someone understand a diagnosis, prepare questions before an appointment or make sense of a letter from the hospital. It is available at anytime and anywhere, and you can ask it about your most embarrassing medical problems from the security of your sofa.
But this also creates risks. AI can sound authoritative and reassuring, even when it is wrong. It can reinforce a misconception, miss an important warning sign or produce information and references that simply do not exist. It does not necessarily know your full medical history and cannot examine you.
There have been deeply concerning reports about people relying on AI for medical advice. In a case reported last week, the parents of a 31-year-old woman with epilepsy brought legal action following her death, alleging that ChatGPT had contributed to her death by advising her to stop taking her anti-epilepsy medication and even providing a schedule for gradually reducing it. These are allegations that need to be tested through the legal process, but the case illustrates how serious the consequences of treating a chatbot as a medical professional can be.
The lesson is not that we should stop using technology or researching our health. It is that we need to become more discerning as the technology becomes more sophisticated.
Learning to recognise misinformation
One interesting approach to tackling misinformation is something called prebunking, in contrast to debunking. Debunking means correcting false information after we have encountered it. Prebunking is about preparing people in advance to recognise misleading information before they come across it.
We can learn to spot some of the common tactics: sensational headlines designed to provoke an emotional response, claims that doctors are hiding a cure, miracle treatments that supposedly work for a multitude of unrelated conditions, or dramatic promises based on very little evidence. There are some simple questions worth asking whenever we encounter health advice. Who has written it? Are they suitably qualified? Is it supported by scientific evidence or based on one person's experience? Are they trying to sell us something? And can the information be checked against a reputable medical source?
When using AI, there are additional considerations. It is reasonable to ask where the information has come from and to check important claims independently.
Some AI tools make it easier than others to check where their information has come from. Some display links to sources alongside their answers, while others may hide these behind a button or only provide them when web searching is enabled. It is worth taking the time to explore these links and check the original information. However, remember that sources may not necessarily be up to date, the presence of a reference does not automatically make an answer accurate, and you cannot assume that the source actually supports what the chatbot is telling us.
We should be cautious about entering personal or identifiable medical information into online tools, and we should never stop medication, change treatment or delay seeking medical help simply because a chatbot has suggested it.
AI in the doctor's consulting room
It is not just patients who are using AI. Healthcare professionals are increasingly exploring how it can support their work.
One example is the use of AI-assisted clinical note-taking software, sometimes called an AI scribe. In our own practice, we have started using Heidi AI to help with the administrative work involved in a GP consultation.
Traditionally, a GP consultation involves listening to a patient while also typing, jotting down notes or trying to remember all the relevant details. This is followed by significant time spent entering the information into the electronic patient record and potentially doing further administrative work, such as writing referral letters or filling in forms. An AI scribe can help by taking notes and structuring the clinical record, leaving the doctor better able to concentrate on the patient. For me, it has been a gamechanger. It makes the consultation feel more like a conversation, as I can maintain eye contact, observe the patient and listen properly rather than constantly looking at a screen.
Of course, it is not infallible. It can misunderstand terminology, miss important details or occasionally include something that was not said. The doctor remains responsible for the accuracy of the clinical record and must always check and edit the notes before approving them. Patients also need reassurance about confidentiality and data protection.
AI is already changing diagnosis and treatment
Beyond the consulting room, AI is being used in a growing number of areas of medicine.
In breast screening, for example, a large-scale trial in Sweden has explored how AI can support the interpretation of mammograms. The findings suggest that AI-assisted screening can help detect more cancers while reducing the workload involved in reading scans, while still keeping radiologists central to the process. AI is also being used to help analyse retinal photographs for signs of diabetic eye disease, identify possible strokes on brain scans and look for patterns in moles that may indicate skin cancer.
There is also considerable potential in medical research. AI can help researchers analyse vast amounts of data, identify patterns and investigate which treatments might work best for particular groups of patients. Drug development may become much faster as a result.
Some of these applications are already being used in clinical practice; others are still developing. But it is very clear that AI has the potential to improve efficiency, support earlier diagnosis and help clinicians make better-informed decisions.
The human element still matters
With all this technology, it is easy to focus on what computers might be able to do. But I think the more important question is what we want them to help us do better.
A GP consultation might seem quite simple on the surface, but it is often about much more than matching symptoms to a diagnosis. When you consult your doctor, they are likely to be weighing up multiple possibilities, assessing risk, noticing subtle clues, bringing your current symptoms together with other aspects of your history, and trying to understand the person behind the symptoms.
As GPs, we often know our patients over many years. We understand their family circumstances, their previous illnesses, their worries and the things that matter most to them. That continuity and trust can make a real difference to the care we provide.
AI can process enormous amounts of information very quickly, but it cannot fully replicate the human relationship between a patient and a doctor. It does not replace empathy, experience or, ultimately, the responsibility of making a clinical decision.
I am so excited about what AI can offer. I am totally in favour of reducing doctors’ administrative burdens, having enhanced tools to support diagnosis and treatment, and benefiting from fast-paced research. I am also keen to give everyone better access to understandable information and to encourage patients to be well-informed and ask questions. But whether the advice comes from a website, an influencer or an AI chatbot, we should retain a healthy degree of scepticism, think critically and know when to seek advice from a real person.
The future of healthcare will undoubtedly involve more AI. The challenge is for both doctors and patients to use it responsibly. Doctors must make sure it enhances the care we provide and that we never lose sight of the important elements that make healthcare human.
Think before you click
When reading health information online or on social media:
Check the source: Who wrote it? Are they suitably qualified?
Look for evidence: Is it based on reliable research or personal anecdote?
Watch for commercial interests: Are they trying to sell you a product?
Be wary of dramatic claims: Miracle cures and sensational headlines should be approached with caution.
Be sceptical: Scrutinise health claims, even when they are presented by someone with medical credentials.
Don’t just follow the crowd: Numbers of “likes” and followers do not prove something is true, and neither do celebrity endorsements.
When using AI chatbots:
Use AI to understand, not diagnose: It can explain medical information but cannot replace a clinician. Use it to prepare for your appointment, not as a replacement for it.
Remember it can be wrong: It may sound convincing even when its answer is inaccurate.
Check important claims: Ask for sources and verify them independently.
Protect your privacy: Think carefully before sharing personal or identifiable medical information.
Check in with your kids: Young people especially may use AI if they have concerns about their physical or mental health. Ask them about it and use it to start conversations.
Don't change your treatment: Never stop medication, alter treatment or delay seeking help based solely on AI advice.



Comments