Using Large Language Models to Do Medical Tasks

In medicine you don't really need to use the generative function of these AI models why would you want to make stuff up like you actually don't. There's a lot of settings you can do to reduce the spontaneity and that's again to do with prompt engineering but more interesting is just using it to tidy up and pass and put the structured data into something which is a useful output. The inflection point for large language models at least was passed last year when we can start teaching these stories to AI in a really useful way.