Australians turning to AI for medical advice, alarming healthcare professionals

In a pharmacy in Mudgee, a small town in regional New South Wales, something has shifted. More customers are walking through the doors of Blooms The Chemist with questions that would have seemed unusual just a few years ago: Is this diagnosis from an AI chatbot actually correct? Should I trust what the algorithm told me?

Pharmacy owner Laura Rheinberger has watched this pattern emerge with growing frequency. The inquiries tend to cluster around two concerns—parents trying to understand what's wrong with their children, and adults wrestling with sleep problems. "We are increasingly seeing people come into the pharmacy after researching their symptoms online or using AI to find out more about how they're feeling and what that could mean," Rheinberger said. A parent might arrive after consulting an AI tool about their child's symptoms. Someone else might show up seeking clarity on a sleep issue they've been trying to decode through a chatbot. Even Rheinberger herself, despite her pharmaceutical training, has turned to AI in the small hours of the morning, searching for ways to soothe her newborn while exhaustion sets in.

The phenomenon extends well beyond pharmacy counters. General practitioners across Australia report the same trend—patients arriving with research already done, armed with algorithmic conclusions about their own health. Dr. Sidd Sarmah, a Queensland-based GP, frames the shift as part of a larger cultural hunger for convenience. "What we look for is convenience, that's what I mean by sugar hits. Anything that's convenient for the patient, they reach out for," he said. The speed and accessibility of AI tools make them attractive, but that convenience carries hidden costs. AI operates through pattern recognition, identifying similarities in vast datasets and offering responses based on what it has learned. Medicine, however, does not always follow predictable patterns. The human body resists neat categorization.

The danger lies in what gets missed when someone bypasses a doctor's assessment. Rheinberger articulated the core limitation: "AI cannot see, listen, touch or feel a patient to gather more information. It cannot order blood tests or conduct imaging. Those important tasks can only be done in person by a healthcare professional." A lump in the neck might be dismissed by an algorithm as a swollen gland from a viral infection—a reassuring conclusion that could mask something far more serious. A rash might fit a dozen different patterns, each with vastly different implications.

Dr. Sarmah recalled a case that crystallizes the risk. A man in his thirties came in complaining of persistent diarrhea. He had traveled extensively for work, and through AI consultation, he had settled on a diagnosis: a stomach bug picked up overseas. He was taking over-the-counter anti-diarrhea medication and assumed his weight loss came from not eating properly during his illness. When Sarmah reviewed his medical history, nothing suggested bowel problems. Basic tests revealed anemia. Further investigation led to a diagnosis of Crohn's disease—a chronic inflammatory condition that requires ongoing medical management. The diarrhea that an algorithm might categorize as gastroenteritis was actually a symptom of something far more complex, something that demanded specialized treatment and long-term care.

What separates a doctor from a chatbot, Sarmah suggested, is not just knowledge but context. "One thing is for sure, and that is that general practitioners will remain at the centre of a patient's healthcare. Wherever the information comes from, whatever the source, we are still the most well-equipped people to interpret that data in the context of a patients needs." A doctor can ask follow-up questions, notice what a patient doesn't say, consider the full arc of someone's medical history, and recognize when a pattern breaks. An algorithm can only match input to output.

Neither Rheinberger nor Sarmah dismiss the value of patients educating themselves about their health. Curiosity and engagement matter. But there is a difference between informed participation in your own care and replacing clinical judgment with algorithmic suggestion. As AI tools become more sophisticated and more ubiquitous, that distinction grows harder to maintain—especially for people without medical training, searching for answers in the middle of the night, desperate for clarity about what their body is telling them. The question facing Australian healthcare is not whether AI will be part of the conversation, but how to help people understand what it can and cannot do.