🤖 AI Summary
A new open-source workflow leveraging large language models (LLMs) has been introduced to enhance the reproducibility and advancement of published economics research. The system operates by utilizing the published replication packages of articles to first reproduce original calculations and identify discrepancies. Out of 4,452 analyzed replication packages across five leading economics journals, the workflow detected inconsistencies in 3,460 articles, significantly highlighting the issues surrounding the reproducibility crisis in economics.
Beyond reproduction, the workflow also improves the efficiency of calculations, with 496 articles benefiting from optimized algorithms that reduce computation time by over tenfold while maintaining accuracy. Furthermore, it extends the original research by proposing new analyses aligned with the original study's objectives, achieving this in 923 articles. This innovative approach not only improves the reliability of economic research but also encourages ongoing development and exploration in the field, marking a crucial step towards integrating AI methodologies within academic research practices.
Loading comments...
login to comment
loading comments...
no comments yet