Three in five Brits now use AI to self-diagnose. That should worry us, and reassure us

There is a particular kind of worry that happens in the middle of the night when a thought enters your head and you can’t get rid of it. Your brain has already decided it needs an answer and quick.

You try to go back to sleep, but instead one thought becomes five and five then becomes 17. Your brain starts to connect the dots that may or even may not belong together. So, what do you do?

You reach for your phone, because your phone is there and the Internet is 24/7. You type a question, then another and that leads to another one and before long, you’re asking a machine to help you make sense of something you can’t quite make sense of yourself.

I am 50, lift weights and go to the gym at least three times a week. I do TRY to look after myself, but I also have a perimenopausal body and brain that sometime disagree with me.

So when something feels odd, I want to know what might be going on. I don’t expect a chatbot to diagnose me, but because my uncertainty is rampant at 3am in the morning, I want to ask without feeling judged, embarrassed or just stupid. And not every human being is someone you want to wake at 3am with a worry and a list of increasingly unlikely possibilities.

That is the question I think gets lost when we talk about AI and health. We debate whether artificial intelligence is safe, accurate or just plain old dangerous. All fair questions, but there is one other thing hiding underneath, and that is why are so many people turning to it at all?

Getting medical help is not always as simple as deciding that you need it.

When I need a GP appointment, I do not necessarily speak to a person first. I now have to type my problem into a screen and work through questions with a Chatbot that are supposed to establish what happens next. It reminds me of those children’s storybooks where you get to choose the ending. Turn to page 35 if you want this outcome; page 68 if you want that one. Except those books were allowed to be unpredictable, and you might make the wrong choice, end up somewhere bizarre and have to start again.

With my health hanging in the balance though, being sent down the wrong path feels rather less amusing. Sometimes the Chatbot works, but sometimes it sends me down a route that does not quite fit what I am trying to explain. Then there is the waiting and the callback that usually arrives while I am on a work Teams call. Then there is uncertainty about whether I should try again tomorrow or see whether the symptom disappears on its own.

And that is where AI becomes strangely attractive. It is there whenever you need it to be. A constant, unjudgemental companion. There is no eye-rolling, no sense that you are taking up too much time, no, “Let’s get the manual out and check what’s really going on before we decide what to prescribe.”

Instead, it gives you an answer in seconds, sounds confident and reassures you where necessary. It appears to understand the problem. So when you are tired, worried or just desperate for someone to tell you what is wrong, it is easy to mistake an instant answer for the right one.

Research cited by Confused.com suggests that AI has already become part of the way millions of people think about their health.

The precise figures matter, but the bigger point matters more: people are already using these tools. The question is no longer whether they should, it’s what happens to the data we are inputting.

Midlife can make health questions messy because symptoms don’t just arrive one at a time. Fatigue could be down to stress, hormones, poor sleep or low iron. There again, it could be a thyroid problem, too much coffee, not enough water, or just being knackered because modern life is so demanding. We can be working, looking after the kids, caring for ageing parents, trying to manage the chores and look after our health at the same time. Something often has to give. While AI doesn’t get rid of any of that, it does give you immediate attention when you need it the most.

There is a temptation to assume that anyone using AI for health information is reckless, but I don’t think that is a fair assumption. Sometimes people are not looking for a diagnosis. They are trying to work out whether something sounds urgent, understand a piece of medical terminology, prepare for an appointment or figure out which questions they need to ask. And, if I am honest, sometimes they are trying to make sure they can get a GP’s undivided attention without being written off in the first five seconds.

For me, the useful part is not being told what I have. It is just having somewhere to start organising the uncertainty.

A chatbot does not examine me, it can’t see how I look, sound or move on any given day and at 3am in the morning, that’s probably a good thing. It doesn’t know my medical history unless I tell it, and can’t perform a physical examination, order the right test or take responsibility for what happens next.

And there is another problem. If you are already anxious about your health, feeding a machine more and more questions can make spiralling worse. One answer produces another question, another symptom appears. Suddenly you are reading about something you had never even considered. False reassurance can also be dangerous and so can unnecessary alarm.

We keep asking whether AI will replace doctors in the future, but I suspect that is the wrong question.

For many people, AI is replacing the gap between having a question and getting an answer. It is replacing the silence between noticing that something is wrong and having the time to discuss it.

So perhaps what it is really replacing is the feeling of sitting alone with uncertainty in the early hours of the morning, when your brain refuses to switch off. That does not make it a doctor, but it does tell us something about the system we have built around us, a system that is supposed to care for us, and one that we already fund collectively through our taxes.

Perhaps the more uncomfortable question is why, having invested so heavily in a healthcare system designed to care for us, so many of us still find ourselves turning to a machine in the middle of the night because it is the only thing that answers.

I think tools like AI can be useful when they help people understand information, prepare questions and decide when to seek proper medical advice. Where I would draw the line though is that I’d never want convenience to become a substitute for care. The responsibility can’t sit entirely with the person holding the phone.

If millions of people are turning to AI because it is faster and easier to access than a healthcare professional, we should really be asking questions about what that says about access to healthcare in the UK. The machine may be the first place we ask the question, but it shouldn’t necessarily be the last place we take the answer from.

People notice change and they want to understand their bodies, look after their families and feel less frightened when something doesn’t feel right. They’re not bad instincts. But we should not confuse the existence of a useful tool with the existence of a good healthcare system.

At 3am, a machine might help me find the words, or give me some form of reassurance. In the morning though, I still want a face to face appointment with a human being who can listen to them and who I trust.

We’re constantly telling our kids off for too much time on their screens, playing too many games. We worry about what happens when they reach for a screen instead of a person. Yet here we are, doing something similar ourselves, only our version feels more grown-up. We’re not asking a machine to entertain us; we’re asking it to reassure us, and sometimes tell us what to do next. So what happens when the first thing we do is turn to a machine?

Midlife, real life, your life.