Navigating Health In The Age Of Misinformation
With the rise of social media and influencers, rapidly changing positions in health policy by governmental institutions such as vaccine policy and now LLMs engines, a perfect storm has emerged for health misinformation, potentially impacting patients getting the right care at the right time. Importantly, it is becoming a significant part of the clinician-patient relationship. My colleague, Dr. Tyler Beauchamp, a pediatric resident at the University of North Carolina, who I co-wrote this piece with, shared this recent experience in the emergency department.
"It was just after midnight when a mother brought her 1-week-old infant into our emergency department. The baby’s only symptom was tachypnea at nearly 80 times a minute. No fever. No rash. A normal pregnancy and delivery. But the child was breathing fast enough to terrify any parent, rightfully so.
“As part of the workup, I recommended we get chest X-ray to look for the cause. The mother adamantly declined, stating ‘No one is going to radiate my baby.’ She explained that she learned about X-rays from the AI tool she had been using. A finger prick to test the baby’s sugar level, was also denied as was staying overnight so we could observe her baby for worsening symptoms. I asked what she would like us to do. She answered a lung ultrasound because the AI tool had told her it would be the safest way to find what’s wrong. When I told her that an ultrasound was not the right test for a high respiratory rate in a newborn, she responded ‘How could you possibly know that?’ ”
Physicians have become all too familiar with this type of conversation, often amplified by social media “experts” and generative AI tools. Most clinicians are led by the Hippocratic Oath in helping the patient with compassion, but what happens when one patient is resistant to quality care and, in the process, affects every other patient who might be waiting for the doctor with their own urgent and emergent issues? Part of our duty is to express patience while helping guide health. With the rise of “Dr. Google” and self-serving AI health, what is the impact to the doctor- patient relationship?
The ER: America’s Primary Care
Across the country, emergency departments are increasingly serving as de facto primary care clinics. According to the Centers for Disease Control and Prevention, millions of visits each year are for non-emergent conditions that could be addressed in outpatient settings. But primary care access is shrinking. Insurance barriers, appointment backlogs, physician shortages and cost concerns have pushed patients toward the one place that cannot legally turn them away — the emergency department.
Emergency physicians now manage hypertension follow-ups at 2 a.m. They refill medications. They evaluate chronic back pain that has been present for years. The ER has become the safety net for a healthcare system that has frayed.
The Rise of ‘Doctor Google,’ Social Media and Distrust
Physicians were once pillars of communities, not because of status, but because they were devoted to protecting their town. People came to them in their greatest need. As medicine has evolved, physicians have had to keep up with an overwhelming amount of information each year, improving their knowledge and commitment to the field and to patients. Yet, the public's perception of what being a physician means has changed, especially over the last decade.
Today, physicians are asked to treat in ways that may be backed by misinformation rather than science. The internet has democratized information, but without a careful guide, information may be conflated with understanding.
With the rise of AI and LLMs such as Chat GPT, a significant problem we may face is being the "yes man" of healthcare. Many of these AI programs are designed to provide responses that align with users’ goals and can portray confidence in their answers. However, those using them briefly and those who do not challenge the tool may receive information that merely echoes their own beliefs.
These systems are not inherently bad. They are tools. A hammer can fix up a home, but it can also end a life. Any AI is only as good as the information as it is given. Where a doctor has been trained to prompt patients to consider all aspects of their illness, a lone AI user will only get answers for questions they ask.
A recent study published in Nature explored outcomes of Chat GPT Health-directed clinical recommendations and found that AI both under triaged over half the cases tested, with the most dangerous failures occurring in emergency conditions. The study highlights not only the risks of missing emergent needs, but the limitations of AI tools not being able to look beyond what is asked of them.
Another part lies in cultural erosion of institutional trust. Over the last decade, misinformation has flourished often due to social media platforms - they live and die by content engagement. Engagement often means showing people content they agree with, accurate or not. Over time, that becomes an echo chamber. If you believe vaccines are harmful and engage with that content, you will be fed content reinforcing that belief. This is why you’ll never see nuanced medical explanations go viral. Our job as physicians is to help our patients make informed decisions, drawing from years of expertise, mastering research, wisdom and human experience.
What happens when medical distrust rises so high that people trust virtual advice over those trained to understand the human complexities AI cannot?AI tools can be extraordinarily sophisticated, but they cannot examine your child. They cannot see subtle retractions in the chest wall or notice that feint crackle. They cannot make a treatment plan that, while not being the “gold standard,” balances risks with your goals to give you safe autonomy. They cannot hold your hand and comfort you when medicine isn’t enough.
And above all, they do not carry moral responsibility for the outcomes they create.
Physicians do.
