There's something undeniably intriguing about the idea of consulting a virtual doctor – a digital superdoctor that's on call 24/7 and promises to diagnose your ailments with a few clicks and keystrokes. It seems like the ideal solution, especially in a time when five million Canadians lack a family physician. 

My initial foray into the realm of online artificial intelligence (AI) doctor websites left me both impressed and bemused. The slick interfaces and intuitive user experiences promised a seamless journey to medical enlightenment, with a veritable smorgasbord of symptom checkers, diagnostic tools and health assessments at my fingertips. Armed with little more than a list of vague complaints and a healthy dose of skepticism, I dove headfirst into the digital abyss.

The first stop was Cody.MD, "an AI doctor for every human." The interface was user-friendly, promising quick and accurate diagnoses in seconds. Dr. Cody introduced itself. “I'm Cody, your AI Co-Doctor, here to help you get better. I've been trained by licensed doctors.” 

I described my typical symptoms after an overnight call shift: headache, sugar cravings, mild nausea, etc. However, I did not get very far with Dr. Cody, as it rudely mentioned, “Please tell me one specific health problem you'd like help with today.”

I plodded on, going back and forth with Dr. Cody. Eventually, Dr. Cody was able to generate a differential diagnosis, which popped up on the screen: "You may have meningitis requiring immediate medical attention."  I wish I could say that my experiences with other AI doctor websites, such as and, were any better. 

They weren't. 

As I punched in my symptoms – a twinge in my left knee, a persistent itch in my right ear, a curious craving for cheese – the virtual doctors sprang into action, churning through algorithms and data points with the efficiency of well-oiled machines. Within moments, my screen was awash with potential diagnoses, ranging from the mundane to the downright bizarre. In my experiments with them, AI doctor websites generated too many false positives and false negatives, rendering them useless as screening tools. 

The use of AI in health care is an evolving field, and there is a lack of standardized regulations or oversight mechanisms governing AI doctor websites. Many of these platforms currently operate without peer-reviewed evidence and clinical validation.

It's easy to dismiss AI doctor websites as gimmicks or novelties, but in time, they will improve. Hemingway once observed that change often happens slowly and then all at once. Change, especially transformative change, is rarely a linear process; rather, it tends to unfold gradually, building momentum over time until reaching a tipping point where it seems to accelerate exponentially.

Perhaps foreshadowing what lies ahead, Google's AI model recently achieved an impressive 85% accuracy rate on a practice test for the US Medical Licensing Examination (USMLE). But for now, human doctors maintain an advantage over our digital counterparts.

Editor’s note: The views, perspectives and opinions in this article are solely the author’s and do not necessarily represent those of the AMA. 

Banner image credit: Gerd Altmann,