Can you trust AI with your health?

Is everyone using AI for health advice now?

AI didn't just appear out of thin air. We’ve been living with versions of it for well over a decade - it’s been quietly powering everything from Netflix recommendations to virtual assistants, acting as a helpful background feature we almost took for granted.

The arrival of ChatGPT in 2022, however, changed the landscape. AI is no longer an invisible helper tucked into apps. It has a voice. When you speak to it, it responds. It can create, converse, and, in health contexts, even present itself in ways that make it sound like an actual clinician.

Dr David Shusterman, a board-certified urologist and Chief Physician at Modern Urologist in New York City, USA, says that compared to even three or four years ago, he sees far more patients with a long list of possible diagnoses they found online or through AI tools.

“Sometimes they’ve read ten different explanations for the same symptom, and many of those explanations contradict each other,” he says. “Instead of coming in with one concern, they’re often overwhelmed and anxious about five or six possible conditions.

“The internet can be helpful for education, but without a clinical context, it can easily turn into information overload.”

Dr David Shusterman

If you rely on AI or algorithm-generated health advice over peer-reviewed data or a qualified professional, remember - you are not speaking to an actual doctor. These tools don't understand the person behind the screen and are no replacement for a professional, individualised assessment.

Shusterman warns that although AI can summarise information, it can’t examine you, review your full medical history in context, or recognise subtle warning signs during a conversation.

“When someone relies exclusively on algorithm-generated advice, important diagnoses can be missed or delayed,” he says.

AI health content often sounds overly confident, creating an illusion of expertise. This is due to a phenomenon called ‘AI hallucinations,’ where the technology generates information that sounds perfectly logical and factual but is actually entirely invented. AI models prioritise flow and persuasion over medical truth, which makes it harder for people to unlearn harmful, oversimplified health advice.

“AI-generated information is often written in a very authoritative tone, which makes it sound like definitive medical guidance,” warns Shusterman. “The issue is that the confidence of the language doesn’t guarantee the accuracy of the information.

“When people hear something stated very confidently online, it can be difficult to convince them that the situation is actually more nuanced.”

Shusterman says the real danger is when safety exceptions are omitted, and people are given one-size-fits-all advice.

“In medicine, small details matter - age, medicines, family history, physical exam findings," he explains. “A recommendation that is safe for one person may be dangerous for another.

“When complex symptoms are reduced to generic advice, you risk overlooking serious conditions that require timely evaluation or specialised treatment.”

Though many people turn to AI for a quick breakdown of health concerns, they often feel compelled to double‑check its answers. This can mean asking the AI follow‑up questions or searching elsewhere, which sometimes creates a merry‑go‑round of contradictions and second‑guessing. Before long, you can spend hours online and end up more confused and worried than when you started.

“Patients sometimes spend weeks or months researching symptoms online, and instead of feeling more informed, they feel exhausted and unsure what to believe,” Shusterman explains. “Eventually, some people delay care because they’re stuck in a cycle of reading conflicting opinions.

“That kind of decision paralysis can, unfortunately, postpone the medical evaluation that would give them clear answers.”

Bypassing credible sources for a quick-fire synopsis from the digital wild west can easily trigger cyberchondria. This is the digital form of hypochondria - excessive worry about having an illness you may not actually have.

Short, bullet-pointed summaries make it easy to skim and miss reassuring context, while highlighting alarming signs. That combination can push you to repeatedly search for confirmations, misinterpret normal sensations as symptoms, and escalate your anxiety.

When AI advice turns out to be wrong - or people rely on unqualified sources, misleading visuals, or deepfake experts - it can undermine trust in real providers and the healthcare system as a whole.

Shusterman says this breeds confusion and scepticism. When people discover supposedly trusted online information is inaccurate, they may start doubting all medical guidance - even advice from real physicians.

“Trust is a critical part of the doctor-patient relationship,” he explains. “Our goal as clinicians is to help people navigate information, not dismiss their curiosity.”