Ask HN: Should primary care doctors be replaced with AI?

By LLM, you mean some flavor of ChatGPT? Have you checked its results using non-LLM sources? It sounds like a mild case of AI mania. Your doctor presumably knows what a dangerous LDL level is, maybe they weren’t concerned because it isn’t dangerous?

I believe it absolutely should be, and it can even be applied to rare disease diagnosis.

My child was just saved by AI. He suffered from persistent seizures, and after visiting three hospitals, none were able to provide an accurate diagnosis. Only when I uploaded all of his medical records to an AI system did it immediately suggest a high suspicion of MOGAD-FLAMES — a condition with an epidemiology of roughly one in ten million.

Subsequent testing confirmed the diagnosis, and with the right treatment, my child recovered rapidly.

For rare diseases, it is impossible to expect every physician to master all the details. But AI excels at this. I believe this may even be the first domain where both doctors and AI can jointly agree that deployment is ready to begin.

I totally get how annoying all this running around is! Having to call back and forth just to get a lab order or a follow-up request, and then the doctor just jumps to conclusions and labels it a "diet issue"—but then the LLM actually asks for details first before giving advice? Anyone would start wondering, “Can AI actually do the doctor’s job?” But honestly, I don’t think it’s about “replacing” them—it’s about “filling in the gaps.” Think about it: what’s LLM good at? Digging through data and picking up on little details, like spotting right away that your LDL doesn’t match your age, even linking it to genetics. Way more accurate than doctors relying on their gut and making assumptions. But what’s it missing? The ability to be there in person to get a feel for things. Like when you see a doctor—they can check if you look pale, if your ankles are swollen, or feel your pulse to tell if your heart rate’s steady. That’s the invisible stuff AI can’t catch. And then there are the tiny symptoms you might not mention—like occasional dizziness or tossing and turning at night—doctors can pull that out just by chatting, but LLM can’t exactly stare through a screen and notice that, right? And you mentioned doctors “only referring people”—but primary care docs are supposed to be the “first filter,” y’know? It’s not just writing orders or sending you to specialists. They should help connect the dots with all your random symptoms. If your LDL’s high, a good one would ask, “Does anyone in your family have this?” or “Do you ever get chest tightness?” Then they’d pair that with your blood work to decide—do we prescribe statins now, or run more tests to rule out other stuff? The doctor you saw just didn’t do that. It’s not that “the primary care role is useless”—she just didn’t do her job as the “gatekeeper.” And about the statins: LLM was right you need them, but the details? How much to take? Do you need to check your liver (since statins can mess with that)? Are you on other meds that might clash with it? Only a doctor can call those shots. I helped my mom with her high blood pressure once—AI suggested a pill that interacted with her diabetes meds. Thank god the doctor caught that! AI can read medical guidelines, but it can’t wrap its head around all your personal stuff the way a human can. And when people say doctors “mess up a lot”—it’s not all their fault. So many primary docs see 30+ patients a day, five minutes each. They don’t have time to ask all the questions, so they default to the “most common cause” (like blaming diet for high LDL). But AI doesn’t care about “saving time”—it’ll list every possibility and ask for clarifications first. That’s why it’s perfect as a helper, not a replacement. Imagine: before the doctor sees you, AI organizes your medical history, flags the weird parts (like your LDL vs. age), then the doctor just focuses on checking those. Saves time and avoids misses. At the end of the day, your frustrating experience isn’t about “do we need primary care docs?” It’s about “how do we stop them from being swamped and careless?” AI can fix the data stuff and the detail questions, but it can’t replace that human call. Taking meds isn’t nothing—you need someone to watch if you get side effects, or calm you down when you’re panicking (“It’s okay, let’s run one more test first”). Right now, only a real doctor can do that.