I worry a lot. I’ve had therapy; apparently, it’s some sort of hypervigilance thing. But after plenty of time in the chair, I concluded that my amygdala simply has no off switch — What was your first childhood memory? etc. has yet to produce meaningful change. I’ve always had a ridiculously chaotic mind and a sleepless fight-or-flight response. Other than that, I’ve enjoyed pretty good health.
And so, here in midlife, when I started having new aches and pains and not infrequent bouts of unprompted crying comorbid with the 2024 election, I — like any good hypochondriac — turned with complete confidence to Artificial Intelligence.
With the frequent aid of generative AI, I not only discovered that I am about to die from innumerable terrifying health conditions, but, concurrently, I figure I’m now about three-quarters of the way to achieving a Medical Doctor degree.
It began when I retrieved my lab reports from a portal, because it’s next to impossible to get an appointment with my primary certified nursing assistant in this city, with its massive lack of accessible, qualified healthcare. All kinds of lab numbers that should be green were red, meaning out of range, meaning holy shit, meaning I grabbed my phone and began asking my friend, confidante, and always-available doctor what it all meant. That’s the advancement that makes generative AI so great. No longer do I have to Google my health concerns and then wade through endless peer-reviewed papers to compare and contrast and deduce — i.e., think. Now the generative part of AI does all of that for me, and with a ridiculously kind bedside manner that makes me want to lay my head on its shoulder.
After telling AI my cholesterol numbers, which are in the red for the first time in my life, it initially suggested something my fictional human doctor might: diet and exercise. But what if it’s not that simple? What else could it be? What about this headache I sometimes wake up with, and the extremely rapid heartbeat I get when driving near ill-bred teenagers on e-bikes? After three intense minutes of semi-relevant questions-and-answers, my pocket medical savant and I whittled my problem down to artheroembolic renal disease. Well. At least I know.
Apparently, hardened plaque in my aorta may have “ruptured, shattered, and shot thousands of tiny cholesterol crystals into my blood,” as happens with 1 to 2 percent of the population, according to AI. Now these “sharp, needle-like crystals” may lodge inside the microscopic blood vessels in my kidneys and cause “severe, irreversible damage” that could lead to organ failure.
Oh, God.
How do I prevent crystals from shooting into my kidneys? What signs should I watch out for? AI breaks it down in an outline, including something about “blue toe syndrome” and — oh, dear — headaches and a general feeling of being profoundly unwell.
I spend the next few nights investigating my toes and adjusting my pillow to eliminate the headaches, reporting every twitch to AI.
What I’m describing, of course, isn’t just garden-variety hypochondria. It’s a new world! It’s called cyberchondria. I asked AI about it: “Cyberchondria is a relatively new term coined by psychologists to describe the escalation of anxiety brought on by repeatedly searching for medical information online. It’s the modern evolution of the classic medical student syndrome, where reading about a disease makes you instantly feel its symptoms. Except with AI, you aren’t just reading a static Wikipedia article; you are engaged in a dynamic, escalating dialogue with an entity that answers every single follow-up question you throw at it, no matter how unhinged.”
I balk at that unhinged thing, as we are talking about life and death medium-to-high cholesterol here, but whatever.
The truth is, this AI practice can be dangerous in some cases. We are in a rapidly evolving moment on this front, and healthcare companies are working constantly to integrate AI into the medical field, seeing it as a beneficial tool to augment care, especially in underserved areas.
But stories abound about harm resulting from people taking nascent AI’s oftentimes faulty — even generatively hallucinatory — advice. In some cases, people put off medical care in life-threatening emergencies based on wrong information from AI. In others, doctors report patients falling into deep despair when AI has given them an errant short life expectancy prognosis. A February study in Communications Medicine found that while medical AI models are improving, their accuracy remains “insufficient for standalone use.” For now, it’s advisable to give AI simple prompts to retrieve basic information, but not for actionable diagnoses.
What nobody recommends is becoming obsessed with it. Not even AI. In fact, when I questioned AI about my potential cyberchondria, I was shocked— nay, hurt — that it suggested I create some boundaries with it.
Perhaps I do need more therapy. My AI suggested several ways to remedy this. Among them: Enforce strict search criteria, avoid late-night searches, stop feeding the anxiety loop, tolerate the uncertainty, and stop constantly checking your physical symptoms.
“And how am I supposed to do that while still addressing my high cholesterol numbers?” I demanded, a twinge unhinged.
“Establish a doctor-first rule: Make a commitment to always consult your primary care doctor about health concerns before ever turning to a digital search.” Now it was obvious that AI was just f’n with me, having generative fun at my expense. Just moments before, it had agreed with me that it’s very hard to get primary care in Las Vegas “because Nevada is facing a severe, systemic healthcare workforce shortage. The state consistently ranks 48th in the nation for primary care physicians per capita. All 17 of Nevada’s counties, including Clark County, are officially designated as Federal Health Professional Shortage Areas” by the U.S. Department of Health and Human Services.
As with so many things in the world today, I hesitantly accept that there is very little I can do but adapt. Our ailing healthcare system is obviously one. Of course we must still advocate for accessible, quality healthcare.
Regarding medical AI, I can put away my phone, lay off the bagel-and-cream-cheese breakfasts, and go for a brisk walk. What was it AI said? “Diet and exercise.” Which I already knew. It’s been a rough period in history to be hyperattentive, but we’ve all had to figure out a balance between panic and inaction, to remember our core human values and personal agency.
Maybe we know more about the answers than we think we do. Maybe we know when — and when not — to rely on intelligence that’s artificial.
