As artificial intelligence tools like ChatGPT, Google Gemini and Grok become increasingly popular, more people are turning to them for quick health advice. But medical experts warn that while these systems can be helpful, relying on them without caution could lead to serious risks.
The appeal of AI-driven health support is easy to understand. Accessing a doctor can be time-consuming, while chatbots offer instant responses at any hour. For many users, these tools feel more interactive and personalized than a standard internet search.
One user, Abi from Manchester, has been using AI chatbots to help manage her health anxiety. She says the experience often feels like having a conversation with a doctor, allowing her to explore symptoms in a less overwhelming way than browsing online, where worst-case scenarios can dominate results.
In one instance, the technology appeared to work well. When Abi suspected she had a urinary tract infection, the chatbot reviewed her symptoms and suggested visiting a pharmacist. She followed the advice and received appropriate treatment.
However, the technology does not always get it right. After a hiking accident left her with severe back pain, Abi consulted the same chatbot and was told she might have a serious internal injury requiring emergency care. She went to hospital, only to later realize the situation was not as critical as suggested.
Her experience reflects a broader concern: AI systems can provide confident answers that are not always accurate.
England’s Chief Medical Officer, Chris Whitty, has warned that while people are increasingly relying on these tools, the quality of responses is still inconsistent. He noted that AI-generated advice can often sound authoritative while being incorrect, creating a potentially dangerous combination.
Research is beginning to reveal why this happens. A study by the University of Oxford examined how chatbots perform in medical scenarios. When given complete and clearly structured information, the systems were highly accurate—correctly identifying appropriate care in about 95% of cases.
But real-life conversations are rarely that precise. When ordinary users interacted with chatbots and provided information gradually or incompletely, accuracy dropped sharply to around 35%. In many cases, people received incorrect diagnoses or inappropriate advice.
Researchers say the issue lies in how humans communicate. People often omit details, misinterpret symptoms or describe them inconsistently, which can lead AI systems to draw the wrong conclusions.
The risks are particularly concerning in serious medical situations. Subtle differences in how symptoms are described can dramatically change the advice given. For example, a severe headache might be interpreted as a minor condition in one case, but flagged as a medical emergency in another—depending entirely on wording.
Medical professionals also highlight a psychological factor. Unlike traditional websites, chatbots create the impression of a personalized interaction. This can make users more likely to trust the information, even when it may not be reliable.
Further studies have raised concerns about misinformation. Researchers in the United States tested several AI systems across topics such as cancer treatments, vaccines and nutrition. More than half of the responses contained some form of problematic or misleading information.
In some cases, chatbots even suggested alternative therapies as viable treatments for serious illnesses—advice that contradicts established medical evidence.
Experts say this stems from how AI systems are designed. Rather than truly understanding medical science, they generate responses based on patterns in language data. This can result in answers that sound convincing but lack clinical accuracy.
Despite these concerns, developers continue to improve the technology. Companies behind major chatbots say they are working with healthcare professionals to enhance safety and reliability. However, they emphasize that these tools are intended for general information, not as a replacement for professional medical care.
For now, most experts agree on a balanced approach. AI chatbots can be useful for basic guidance, understanding symptoms, or deciding whether to seek help. But they should never be the sole source of medical advice—especially in urgent or complex situations.
Users are encouraged to verify information with trusted sources and consult qualified healthcare providers when necessary.
Abi, reflecting on her own experience, continues to use AI tools but with caution. She advises others to treat chatbot responses as guidance rather than fact, and to remain aware that mistakes are possible.
As artificial intelligence becomes more integrated into everyday life, its role in healthcare is likely to grow. But until accuracy improves significantly, the message from experts remains clear: convenience should not replace professional judgment when it comes to your health.
0 Comments