New Review clarifies how to validate LLMs used as human proxies
A Nature Computational Science Review by Nikita Karetnikov, Iyad Rahwan, and Davor Svetinovic separates four uses of LLMs as human proxies and matches each to a different validity question.
Read the update