Human doctors take years to train, and the resources to train enough are so limited that few countries have enough doctors. We are so used to that state of affairs, it's hard to imagine having a magic wand that could be waved to solve the problem overnight.
Yet, that is almost what AI can do. According to the World Bank’s 2025 Global Findex Digital Connectivity Tracker, about 68 % of adults in developing (low- and middle-income) economies own a smartphone. That means almost everyone has access to one they own, or someone close to them owns. Smartphones are a perfect way to access this AI.
As soon as 2030, everyone on the planet, even the very poorest, will have access to expert medical advice. This should start to feed through to dramatic improvements in health statistics, child mortality, and lifespan improvements.
Cheap AI chatbots transform medical diagnoses in places with limited care: Studies in Rwanda and Pakistan reveal real-world utility of chatbots in underfunded clinics, and not just in benchmark tests.
There's really strong diminishing returns in healthcare. Improve hygiene, chop away obvious cancer, a dozen other simple types of surgery, and basic meds in obvious cases? Stuff like that probably gets you 60% to where contemporary best practice is, for 5% of the cost. Autonomous LLM doctors would fit in great here.
And in developed countries, a medically trained LLM as a "second opinion" is IMHO an excellent use case of the technology. Provided the doctors managed to maintain medical thinking, keep up to date with their own knowledge, and bother to look into patient histories to the same degree as before. Which is a big if, they're only human too.