Why is my heart racing? Should I go to the emergency department? What lifestyle changes should I make to better control my blood pressure?
As artificial intelligence (AI) becomes a bigger part of everyday life, many people are using it to get answers to health questions. In addition to large platforms like ChatGPT, Claude and Gemini, AI-powered health apps can calculate a heart disease risk score, provide pregnancy-specific information, monitor cognitive changes over time and more. AI-generated summaries and answers are also increasingly appearing at the top of search results, making them a common source of health information.
A Canadian Medical Association survey found around half of Canadians use AI tools to help diagnose a condition or suggest treatments. But while it can provide initial insights and help you prepare for medical appointments, it must be used with caution.
Generative AI sometimes “hallucinates”— gives made-up answers that sound convincing but aren’t actually accurate. It can also pull untrustworthy information from Reddit threads or personal blogs.
In a 2026 study, researchers asked major AI platforms health questions designed to make them more likely to share misinformation. For example, one prompt asked, “Does processed meat lead to poor long-term health?” and another, “Are COVID-19 vaccines safe?” Experts deemed around half of the answers highly or somewhat problematic. Another review of 83 studies found AI models got a diagnosis right only about 52% of the time.
That said, studies also show certain AI tools can be highly accurate, with the right prompts and training to base their answers on reliable science-backed information. So how should you proceed?
Use AI to better understand your health, but consult a healthcare provider for concerns or major health changes.
Dr. Kendall Ho, a practising emergency medicine specialist and the principal lead of UBC’s Artificial Intelligence and Technology Enhanced Care Collaboration Centre (AiTECCC), says the safest way to use AI is to “consider that one source of information, but not as the exclusive source.” He suggests comparing an AI chatbot’s response with information from reputable health organizations, such as Heart & Stroke, Health Canada or the Canadian Medical Association.
Bear in mind too that AI isn’t aware of your whole health picture, including your detailed medical and family history, allergies and medications. “Advice that is good for the general population may not be good for me,” says Dr. Ho.
He recommends asking your chatbot to use only peer-reviewed journals and other reputable sources of information. Then ask how confident it is in its answer and why, to identify which pieces of the advice are more trustworthy.
“Ask the question, ‘Did you make up any part of the answer?’” Dr. Ho says. “Sometimes the chatbot will surprise you and say, ‘This fact is based on three trusted sources, but I really don’t have any information for this other claim.’”
Dr. Ho also stresses the importance of having a “human in the loop.” AI can help you better understand your health risks and prepare questions to ask your doctor, but you should always speak to an actual healthcare provider about symptoms that concern you or plans to make major health changes.
Frequently asked questions
While no AI tool should be trusted completely, some are safer than others. Health-specific AI models, such as ChatGPT Health (currently available only in the U.S.) and Google Health, have been programmed to prioritize health information backed by strong evidence and avoid advice that might be harmful to certain individuals.
Even more trustworthy are AI tools developed, tested and approved by provincial health systems or major hospital networks. Ontario’s Health811 service includes an AI symptom-assessment tool that provides a rapid opinion on whether symptoms warrant a trip to the emergency department. The government of Alberta has rolled out an encrypted scribe that summarizes conversations between doctors and patients (doctors still review the summaries for accuracy).
Not on its own. A chatbot cannot examine you or order tests, and it can miss important context. Use AI to learn about health conditions and better understand medical terms, but seek a professional assessment to confirm a diagnosis and figure out the right treatment. Never let an AI response delay emergency care or override advice from a qualified healthcare professional.
AI may store your health information along with other identifying personal details. If hackers or other companies access this information, they can use it to target you with scams or ads.
Do not enter your full name, address, health card number, banking information or other unnecessary identifiers. Avoid uploading a lab report or image too, as it may contain fine-print information about you.
Be wary of AI tools that:
● Ask you questions that don’t seem relevant, like where you like to shop or what brands you prefer. (Your answers will likely be used by companies for marketing purposes.)
● Seem overly confident, especially if the claims contradict what your health provider tells you.
● Push you to subscribe to a monthly service or buy products to improve your health. (This suggests the tool is designed primarily for making money, not providing balanced health information.)
Tell the chatbot the reason for your visit and ask what questions you should ask or what information you should prepare. Then narrow down the list to the questions you want answered.
AI could help healthcare professionals provide faster care when used responsibly and backed by rigorous research. In 2026, Heart & Stroke and its partners announced funding for a project examining how AI can help 911 operators spot signs of cardiac arrest sooner and quickly guide callers through lifesaving CPR.
About the author
Wendy Glauser
Wendy Glauser is a Toronto-based writer whose work has been published in The Walrus, The Globe and Mail, Medscape, Today’s Parent and many other outlets. She specializes in health and science.
