I can't go back to edit, but I understand the downvote. It seems like I am saying only LLMs will be the sole authority. And that doesn't make sense, insofar as

I can't go back to edit, but I understand the downvote. It seems like I am saying only LLMs will be the sole authority. And that doesn't make sense, insofar as a model without the latest data may make the wrong recommendation, and--more importantly--the doctor would have the final say in the treatment plan. What I mean is, if LLMs are able to match "here is what any reasonable person of equal experience would have done in the circumstances," [1] it would be very risky to try something counter to that. The Large Language Model as Liability Limiting Machine. [1] https://himpro.ca/reasonable-care-vs-standard-of-care/ https://himpro.ca/reasonable-care-vs-standard-of-care/ > Reasonable care is the caution and concern used by an average, rational person in any given situation to prevent the harm of others. > Standard of care is the acceptable standard of care acceptable in society or in a profession.