Automated reproducibility assessment of social-science research with LLMs

arXiv·medium signal

This work uses large language models to automate reproducibility evaluation in the social and behavioral sciences — work normally done by independent researchers reanalyzing original datasets. It is a concrete applied-LLM workflow for research QA and a template for builders applying LLMs to verification/audit tasks rather than generation.