**
In a concerning trend, a recent report reveals that a significant number of Canadians are increasingly seeking medical guidance from AI chatbots, with troubling results. As reliance on these digital assistants grows, a study conducted by the Canadian Medical Association underscores the potential dangers of this shift. While AI technologies like ChatGPT are now commonplace, their ability to provide accurate medical information is falling short, leading to adverse health outcomes for those who trust them.
Rising Popularity of AI in Healthcare
For years, patients have taken to the internet to research medical symptoms and seek self-diagnosis. However, the emergence of advanced AI chatbots has transformed this landscape. According to OpenAI, around 40 million individuals globally utilise ChatGPT for health-related inquiries each day. In Canada, half of the respondents in the Canadian Medical Association’s latest survey admitted to consulting AI tools like ChatGPT or Google AI for health concerns.
The reliance on these technologies has proven problematic. Those who adhered to AI-generated medical advice were found to be five times more likely to experience negative health effects compared to individuals who sought traditional medical consultations. The reasons for this are concerningly clear. AI systems often exhibit overconfidence and a tendency to provide misleading information, which can lead to misguided treatment choices.
The Flaws of AI Diagnostics
Research from the University of Waterloo highlights the shortcomings of GPT-4 when faced with medical questions, revealing that it delivered incorrect responses approximately two-thirds of the time. Similarly, a study conducted by Harvard researchers demonstrated that chatbots frequently fail to challenge nonsensical queries, such as confusing acetaminophen with Tylenol, despite being the same medication.
Moreover, Google’s AI Overviews, which are now ubiquitous in search results, are not much better. An investigation by The Guardian recently revealed that these summaries often contain erroneous information, prompting expert responses that describe the content as “alarming” and “dangerous.” Notably, one instance incorrectly advised pancreatic cancer patients to avoid high-fat foods, while another inaccurately suggested that pap smears were necessary for vaginal cancer screening.
The Influence of Non-Expert Sources
Further compounding the issue, Google’s AI Overviews have increasingly cited YouTube as a source of health information. A study examining 50,000 Google health queries found that nearly 4.5 per cent of AI Overviews referenced YouTube content, outpacing citations from recognised medical institutions and government health sites. This trend raises significant concerns, as YouTube is not a reliable medical source, and anyone can upload content to the platform.
Despite the warnings present in AI companies’ terms of service—stating that their platforms do not replace professional medical advice—these disclaimers have become less frequent. In 2022, over 26 per cent of AI responses included disclaimers about their qualifications. By last year, that percentage had plummeted to below 1 per cent.
A Gamble in the Absence of Care
Canadians are not blind to the limitations of AI healthcare solutions. The CMA survey indicated that only 25 per cent of respondents trust these platforms for accurate health information. Yet, with challenges such as long wait times for specialists and a shortage of family doctors, many find themselves turning to chatbots as a last resort. CMA president Margot Burnell acknowledged this dilemma, stating, “If you don’t have ready access to care, this is where you go.”
Why it Matters
The growing reliance on AI chatbots for medical advice represents a significant shift in how Canadians approach their health. While these technologies may offer immediate answers, the potential for harmful misinformation poses serious risks. As the healthcare landscape continues to evolve, it is imperative that individuals remain vigilant, prioritising verified medical sources and expert guidance over the convenience of AI. The current trends call for a critical reevaluation of how we integrate technology into our health decisions, ensuring that the quest for quick answers does not compromise our wellbeing.
