In some fields like comp sci, when code isn't given but the paper describes the approach, LLMs do help with the reproducibility crisis: you can ask it to reprod

In some fields like comp sci, when code isn't given but the paper describes the approach, LLMs do help with the reproducibility crisis: you can ask it to reproduce the result through reimplementation by reading the paper. If it fails you may have to double check it did properly reimplement it, but if it succeeds you do get a reproduction.