Hm, this is an interesting blueprint for building an investable regulatory pathway. But it treats healthcare as a billing system to be optimized, as opposed to a scientific domain to be transformed. Frontier AI will reshape diagnosis, prevention, and our models of disease more than it will reshape medical billing codes. The real healthcare revolution will not come from anything in this piece, but from rethinking biology.
These tools require investment of capital, business visionaries, extensive IT investment, partnership from caregivers skeptical about promoting a technology that will in some cases supplant or replace them, buy-in from hospitals just scraping by, momentum to overcome lobbyists invested in the status quo, insurance company support, and a countless number of other vested interests. If you think those forces will usher these changes in as unpaid volunteers, you misread the healthcare industry. I have seen promising healthcare technologies wither on the vine because their development threatened cash cows. Focusing on billing is the EXACT correct way to incentivize development and adoption.
The analysis provides a strong case for AI's potential in healthcare but appears to understate the complexities of current healthcare cost structures. The perspective leans toward viewing AI as a panacea, which overlooks the necessary systemic changes required for effective and meaningful implementation.
Healthcare is one use case for AI technology where the value will actually exceed the current hype. It is hard to explain how much potentially useful medical information is silo’ed or hard or just unprofitable to incorporate into medical decision making, how effective models can be in differential diagnosis, and how models will revolutionize everything from image-based diagnosis to physical exam. AI will be a huge cost saver, force multiplier, and lead to much more effective, cost sustainable, and actually accessible/timely care, allowing the humans involved in delivering care to focus on the human side. This is the “low hanging fruit” of profitable AI deployment.
I lived in Turkey from 2020-2024, and was very impressed by their digital health system.
My wife can open her e-devlet ("e-state") app and see the exact results of a lab test from over 15 years ago (measurements of dozens of variables). When she got imaging done, everything just works for her other doctors to see it, with no thought of "in-network" or any kind of system.
Someone who worked at the US Embassy in Ankara told me, "this is why Erdoğan won" when talking about it.
There certainly is a trade-off of privacy. And the notion of a permanent record can be a bit scary - there are many real examples where it can make people rightfully feel nervous. One of my wife's peers had his pilot career derailed because of a test result from years before. But when the data is needed, especially in an emergency situation, it is at least there and accessible by doctors.
Great article — and the point at the end about artificial constraints on expanding care delivery is critical. These barriers are exactly what keep healthcare costs high. This is fundamentally a business-model challenge, not a technology challenge.
The resistance is often framed as concern over safety or regulation, and it shows up through laws, licensing limits, and reimbursement restrictions. But the underlying dynamic is consistent: the same groups that restrict physician supply will view AI — and its productivity unlock — as a threat. Increased efficiency is deflationary to healthcare costs and therefore to their income, creating strong incentives to slow or block it.
Perhaps productivity gains could eventually allow incomes to hold steady, but with tightly limited physician supply, there’s no competitive pressure forcing that evolution. We’re starting to see cracks in the model (cash-pay markets, etc.), but the real unlock would be dramatically increasing the number of physicians. With more supply, new reimbursement models could emerge and productivity valued.
85+% of all health issues are chronic, and are poorly dealt with thru the current health systems (including many doctors). The solution lies primarily outside healthcare and in the realm of wellthcare. That solution is AI needy, and a much larger ITAM than the nicely written article above.
Hm, this is an interesting blueprint for building an investable regulatory pathway. But it treats healthcare as a billing system to be optimized, as opposed to a scientific domain to be transformed. Frontier AI will reshape diagnosis, prevention, and our models of disease more than it will reshape medical billing codes. The real healthcare revolution will not come from anything in this piece, but from rethinking biology.
These tools require investment of capital, business visionaries, extensive IT investment, partnership from caregivers skeptical about promoting a technology that will in some cases supplant or replace them, buy-in from hospitals just scraping by, momentum to overcome lobbyists invested in the status quo, insurance company support, and a countless number of other vested interests. If you think those forces will usher these changes in as unpaid volunteers, you misread the healthcare industry. I have seen promising healthcare technologies wither on the vine because their development threatened cash cows. Focusing on billing is the EXACT correct way to incentivize development and adoption.
Why can’t we find a way to make AI healthcare circumnavigate the traditional insurance process rather than force it to conform?
The analysis provides a strong case for AI's potential in healthcare but appears to understate the complexities of current healthcare cost structures. The perspective leans toward viewing AI as a panacea, which overlooks the necessary systemic changes required for effective and meaningful implementation.
Healthcare is one use case for AI technology where the value will actually exceed the current hype. It is hard to explain how much potentially useful medical information is silo’ed or hard or just unprofitable to incorporate into medical decision making, how effective models can be in differential diagnosis, and how models will revolutionize everything from image-based diagnosis to physical exam. AI will be a huge cost saver, force multiplier, and lead to much more effective, cost sustainable, and actually accessible/timely care, allowing the humans involved in delivering care to focus on the human side. This is the “low hanging fruit” of profitable AI deployment.
Sure…
I lived in Turkey from 2020-2024, and was very impressed by their digital health system.
My wife can open her e-devlet ("e-state") app and see the exact results of a lab test from over 15 years ago (measurements of dozens of variables). When she got imaging done, everything just works for her other doctors to see it, with no thought of "in-network" or any kind of system.
Someone who worked at the US Embassy in Ankara told me, "this is why Erdoğan won" when talking about it.
There certainly is a trade-off of privacy. And the notion of a permanent record can be a bit scary - there are many real examples where it can make people rightfully feel nervous. One of my wife's peers had his pilot career derailed because of a test result from years before. But when the data is needed, especially in an emergency situation, it is at least there and accessible by doctors.
Great article — and the point at the end about artificial constraints on expanding care delivery is critical. These barriers are exactly what keep healthcare costs high. This is fundamentally a business-model challenge, not a technology challenge.
The resistance is often framed as concern over safety or regulation, and it shows up through laws, licensing limits, and reimbursement restrictions. But the underlying dynamic is consistent: the same groups that restrict physician supply will view AI — and its productivity unlock — as a threat. Increased efficiency is deflationary to healthcare costs and therefore to their income, creating strong incentives to slow or block it.
Perhaps productivity gains could eventually allow incomes to hold steady, but with tightly limited physician supply, there’s no competitive pressure forcing that evolution. We’re starting to see cracks in the model (cash-pay markets, etc.), but the real unlock would be dramatically increasing the number of physicians. With more supply, new reimbursement models could emerge and productivity valued.
85+% of all health issues are chronic, and are poorly dealt with thru the current health systems (including many doctors). The solution lies primarily outside healthcare and in the realm of wellthcare. That solution is AI needy, and a much larger ITAM than the nicely written article above.
Typo: but if it relates to doctoring it’s healthcare AI our book. (Should be "in our book")