Don’t let Google or AI diagnose you before you see a health professional
Monday, September 28, 2026
Doctors perform surgery at King Faisal Hospital. Photo by Craish Bahizi.

More people are turning to Google, TikTok and artificial intelligence tools to find out what is wrong with them before seeing a health worker. A fever, headache or persistent cough can now send someone to a search engine or chatbot before they visit a health facility.

The attraction is understandable. Health information is available on a phone at any hour, at no cost and without a queue. Visiting a health facility may involve transport costs, time away from work or class, and waiting.

Studies in Rwanda show a similar pattern. Among University of Rwanda students who treated themselves with antibiotics, half said the illness did not seem serious enough to see a health worker.

Others relied on what they knew or said lectures left little time for consultation. Among parents of young children, close to 78 per cent reported giving medicine without consulting a health professional.

Searching for health information can be useful. The danger begins when an online search becomes a diagnosis and is used to decide which medicine to take.

A person may read that symptoms are "probably nothing” and delay care, allowing a treatable illness to worsen. The opposite can happen too: a simple headache may lead someone to information about brain tumours. The disease may not be there, but the fear can be real.

There is also the risk of taking the wrong medicine. Drugs taken without proper assessment may hide symptoms, cause side effects or interact badly with other medicines.

Antibiotics are a good example. Among University of Kigali students, coughs, colds and flu were common reasons for self-treatment with antibiotics, with amoxicillin the most commonly used.

Yet most colds and flu are caused by viruses, which antibiotics cannot treat. Unnecessary use contributes to antimicrobial resistance, making medicines less effective when needed.

Symptoms can also be misleading because many illnesses share similar signs.

A fever in Rwanda could result from malaria, typhoid, a urinary infection, pneumonia or flu. Weight loss and poor appetite may be linked to diabetes, tuberculosis, HIV, thyroid problems, cancer or depression. Headaches, weakness and joint pain can occur in many diseases.

This is why proper assessment matters. A health worker can ask questions, examine you and order tests where necessary, including a rapid malaria test. A screen cannot examine your body or assess your condition in context.

Research has shown the limits of digital diagnosis. In one study of 23 online symptom checkers, the correct diagnosis appeared first in only 34 per cent of cases.

Newer AI chatbots are more sophisticated, but ordinary users may still struggle to reach the right conclusion. A 2026 study in Nature Medicine found that chatbots working alone identified the correct conditions in about 95 per cent of cases. Yet people using them identified relevant conditions in fewer than 35 per cent.

Researchers found that chatbots could mix good advice with poor advice, leaving users to decide what applied to them.

Online health information still has value. AI and online sources can help people learn about prevention and healthy living, prepare questions before a consultation, or better understand a diagnosis already given by a professional.

Knowing your body is also important. Breast awareness helps people notice unusual lumps, skin changes or nipple discharge and seek medical advice. The same applies to changes in the testicles or moles.

But online tools should not be used to name your disease or decide which medicine to take.

Some symptoms should not send you searching online. Chest pain, difficulty breathing, sudden weakness on one side of the body, fainting, or a high fever in a child should prompt medical attention.

The author is a medical doctor and a passionate public health advocate.