AI models are making paywalled scientific research invisible by only citing and learning from articles that are free to read.
Language models amplify the visibility gap in science by accelerating the citation rates of Open Access papers over others. Because AI primarily ingests data that is not behind a paywall, it effectively erases millions of high-quality studies from the digital record. Researchers often believe that the quality of their work is the most important factor in how often it is cited. This audit proves that accessibility is now the dominant force in determining which scientific ideas survive in the age of AI. Paywalled research is being systematically filtered out of the global scientific consciousness.
Do LLMs Read for Free? Auditing the Open Access Bias in AI-Mediated Science
SocArXiv · xym8d_v1
How large language models affect various social activities has been widely discussed; however, how large language models themselves influence the mechanisms of scientific knowledge production remains unclear. To investigate this question, this paper designs a natural experiment based on a large-scale bibliometric dataset. Using a Difference-in-Differences model with Poisson Pseudo-Maximum Likelihood estimation and analyzing citation records from 64,841 articles across 120 communication journals,