AI is becoming our new health coach – but when should you NOT listen to it?

Need a healthier dinner using whatever is left in the fridge? AI can give you ten ideas in seconds. Want a beginner’s walking plan, some stretches to break up a day at your desk or suggestions for getting more fibre into your diet? No problem. Increasingly, people are turning to artificial intelligence for exactly this kind of everyday health and wellbeing advice – essentially putting a health coach in their pocket that is available 24 hours a day and never gets bored of being asked another question.

There are some genuinely useful things about that. AI can help people understand complicated health information in simpler language, generate ideas, organise routines and prepare questions to ask a healthcare professional. But there’s an important catch: an AI assistant can produce an answer that sounds extremely confident, detailed and convincing while still being wrong.

And when we’re talking about your health, that distinction matters.

Where AI can genuinely be useful

Think of AI as an assistant rather than a clinician and it starts to make much more sense.

It can be great for relatively low-risk tasks. You might ask for a week’s worth of lunch ideas containing more vegetables, ways to add movement into a sedentary working day, a beginner’s strength-training schedule or ideas for improving your bedtime routine. You could ask it to explain an unfamiliar medical term you’ve encountered, turn NHS guidance into plain English or help you create a list of questions before a GP appointment.

It can also be useful for motivation and organisation. If you’ve decided you’d like to walk more, for example, AI could help you build a gradual plan around your working hours and suggest ways of keeping track of progress. None of this requires AI to diagnose you or make a clinical decision.

There is still a sensible rule to remember: when information matters, check where it came from. For health information in the UK, sources such as the NHS, NICE, the Medicines and Healthcare products Regulatory Agency and other recognised medical organisations are considerably more reassuring than an answer with no identifiable source behind it.

Where the line starts getting dangerous

Problems arise when we stop asking AI for information and start asking it to make medical decisions.

Imagine someone develops chest pain and asks an AI assistant whether they need urgent medical attention. Or someone uploads blood-test results and asks whether they can rule out cancer. Perhaps a person taking several medicines asks whether it’s safe to stop one because they’re experiencing side effects.

Those aren’t really “wellbeing” questions anymore. They’re clinical decisions where getting the answer wrong could have serious consequences.

AI doesn’t examine you. It cannot reliably assess all the things a healthcare professional may consider when evaluating symptoms, and the information it receives is only as complete as whatever you’ve typed into the box. A seemingly insignificant detail you haven’t mentioned could completely change the clinical picture.

AI can also generate inaccurate information – sometimes referred to as a “hallucination” – including plausible-sounding explanations or even sources that don’t exist. The confident tone of an answer shouldn’t therefore be mistaken for evidence that it is correct.

For symptoms that could represent an emergency, AI shouldn’t become an additional hurdle between you and appropriate medical help. In the UK, urgent and emergency services such as NHS 111 and 999 exist specifically to assess situations where timing can matter.

Be particularly careful with medication

Medication is another area where the stakes become considerably higher.

AI may be able to explain what a medicine is generally used for or help you understand terminology contained in official patient information. What it shouldn’t replace is personalised advice from a doctor, pharmacist or other qualified healthcare professional about whether you should start, stop or change a treatment.

That includes prescription medicines, weight-loss treatments, antidepressants, blood-pressure medication and seemingly routine medicines that may interact with other treatments or health conditions.

If an AI response appears to contradict advice you’ve received from the professional responsible for your care, don’t quietly change your medication based on the chatbot. Raise the question with your clinician or pharmacist.

Think before sharing your entire medical history

There’s another part of using AI for health that receives considerably less attention: privacy.

When you’re chatting to a friendly-looking box on a screen, it’s surprisingly easy to forget that you’re entering personal information into a digital service. Before uploading medical letters, photographs, test results or details about your health, consider what information you’re sharing and check the privacy settings and policies of the particular service you’re using.

You may not need to provide your full name, address, NHS number, date of birth or other identifying information simply to ask a general health question.

The same caution applies at work. Healthcare and social care professionals should never paste identifiable patient or service-user information into a public AI system unless their organisation has specifically approved the system and its use for that purpose.

Use AI to help you think – not to make the decision for you

AI is likely to become an increasingly normal part of how we manage everyday life, and health will inevitably be part of that. Used sensibly, it can make reliable information easier to understand, help us organise healthier habits and give us ideas we might not have considered.

The answer isn’t to be frightened of it.

It’s to understand what it’s good at – and what it isn’t.

Ask AI to suggest five easy lunches. Ask it to build a walking schedule around your shifts. Ask it to explain what cholesterol means before your health check. Ask it to help you prepare questions for your doctor.

But when the question becomes “What disease do I have?”, “Is this serious?”, “Can I stop taking this?” or “Do I need emergency treatment?”, it’s time to stop treating your phone like a doctor.

AI can be a very useful health assistant. It just shouldn’t be the person in charge of your healthcare.

Medical Disclaimer: The information provided by Daily Round is for general informational and educational purposes only and should not be considered medical advice, diagnosis or treatment. Artificial intelligence tools should not replace assessment or advice from an appropriately qualified healthcare professional. If you believe you may be experiencing a medical emergency in the UK, call 999 or attend an appropriate emergency service.

Stay informed with the stories shaping health and social care.