I think 90% of everyone’s needs are handled by primary care (family med, IM, peds) - most people aren’t going for anything really complicated, and basic preventive care like a yearly physical can be really, really powerful. I think there’s a movement to move away from needing MDs for primary care, which should generally help with availability and costs, as well, but I do think the majority of people don’t need specialty care.
I think adding AI makes sense, though the biggest focus right now (for major providers, at least) is on admit prevention (and, specifically, readmit prevention) and reading imaging. I imagine eventually you could get to quality diagnosing/suggesting, but I think it’s further away than the tech sector thinks. That said, I who knows, if an AI stops hallucinating, maybe you can get closer faster.
The biggest issue in the US, in my opinion, is that fee-for-service models are absolutely, completely broken. They (and bad reimbursement rates) encourage doctors to see too many patients, sell too many non-medical services (see how many ads your dermatologist has for Botox, for instance), and suggest too many irrelevant tests (radiology for lower back pain, for instance). You really need a massive structural change to convince everyone to move to value-based models (which also aren’t perfect, but at least align incentives).
This is a long comment (leaving a lot of space for people to nitpick and/or disagree!), but I’d just end with: healthcare is really, really complicated. Anyone selling an Easy Solution is either a neophyte (I don’t say negatively), or a huckster. What might work in SF will probably work very differently in West Texas or Eastern Montana.