Rigor and correctness of LLM-generated astrophysics

Establish whether astrophysics projects conceived, executed, and written by large language models can meet accepted standards of methodological rigor and correctness.

Background

In evaluating the extreme policy of fully embracing LLMs to conduct astrophysics end-to-end, the author notes several potential alignments with core scientific values but highlights a key uncertainty about rigor and correctness.

The concern reflects current difficulties in assessing LLM-generated content and the potential mismatch between the scale of automated output and human capacity for verification.