LLM Use Acknowledged in Biology Preprints Nears Specialized AI Tool Rate at 17%
Data from Epoch AI shows that the share of biology preprints acknowledging large language model use has risen to 17%, nearly matching the 19% rate for specialized AI tools.
Research organization Epoch AI released a data insight on October 30, 2026, showing that the share of papers on bioRxiv (a preprint server for biology) acknowledging the use of Large Language Models (LLMs) has grown exponentially over the past three years. The rate now stands at 17%, nearly catching up to the 19% share of papers acknowledging the use of specialized AI tools such as AlphaFold or image analysis software.
This data is based on an automated scan of all bioRxiv preprints from January 2020 to September 2026. Epoch AI notes that while specialized AI tool usage has climbed steadily since 2021, driven largely by AlphaFold's popularity in protein structure prediction, LLM usage surged after 2023. The applications for LLMs are more varied, including literature review, writing assistance, and code generation.
It is worth noting that while overall LLM usage approaches that of specialized tools, the proportion of papers explicitly acknowledging LLM use for 'research or analysis' core tasks remains lower, at about 4%. This suggests that most researchers currently view LLMs as auxiliary tools rather than direct engines of scientific discovery.
This trend reflects the rapid penetration of general-purpose AI into professional scientific fields, marking a shift from reliance on specific functional software to the integration of general intelligent assistants in research workflows.