This article also shows how LLMs are able to perform medical labeling tasks at similar accuracy to human annotators through zero-shot prompting.

Academia tends to lag a year or two behind the times, but I believe this is a valuable and extremely practical way LLMs should be used in practice.