Student or educator? Sign up with your school email and get 6 months of Pro free.
Cited by 82scientometrics and bibliometrics researchAcademic Writing and PublishingAcademic Publishing and Open Access

The DOI matches its Crossref record; 42 of the 42 references it cites were found in a registry or an index.

Quantifying large language model usage in scientific papers

Abstract

Scientific publishing is the primary means of disseminating research findings. There has been speculation about how extensively large language models (LLMs) are being used in academic writing. Here we conduct a systematic analysis across 1,121,912 preprints and published papers from January 2020 to September 2024 on arXiv, bioRxiv and Nature portfolio journals, using a population-level framework based on word frequency shifts to estimate the prevalence of LLM-modified content over time. Our findings suggest a steady increase in LLM usage, with the largest and fastest growth estimated for computer science papers (up to 22%). By comparison, mathematics papers and the Nature portfolio showed lower evidence of LLM modification (up to 9%). LLM modification estimates were higher among papers from first authors who post preprints more frequently, papers in more crowded research areas and papers of shorter lengths. Our findings suggest that LLMs are being broadly used in scientific writing.

Overview

Integrity signals

Registry

Crossref · 10.1038/s41562-025-02273-8

Cites

42 of 42 references verified0 citation errors · 0 not checkable

Statistics

Recomputing p-values needs the paper’s full text.

Verification

Claims the paper makes

This paper’s full text isn’t openly available, so its claims and self-check weren’t run. The registry and citation checks still apply.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Citations

Cites

Show all 38 references

Research works printed but not linked: 15

  1. Yu, S., Luo, M., Madasu, A., Lal, V. & Howard, P. Is your paper being reviewed by an LLM? Investigating AI text detectability in peer review. In NeurIPS Safe Generative AI Workshop https://openreview.net/forum?id=f2G7C2fKxV (2024).
  2. Clarification on large language model policy LLM. ICML https://icml.cc/Conferences/2023/llm-policy (2023).
  3. Wiley. ExplanAItions: an artificial intelligence study by Wiley. https://www.wiley.com/en-us/ai-study (2023).
  4. Lavergne, T., Urvoy, T. & Yvon, F. Detecting fake content with relative entropy scoring. Pan https://dl.acm.org/doi/10.5555/3053718.3053722 (2008). · doi:10.5555/3053718.3053722
  5. Badaskar, S., Agarwal, S. & Arora, S. Identifying real or fake articles: towards better language modeling. In International Joint Conference on Natural Language Processing https://aclanthology.org/I08-2115/ (2008).
  6. Mitchell, E., Lee, Y., Khazatsky, A., Manning, C. D. & Finn, C. DetectGPT: zero-shot machine-generated text detection using probability curvature. In Proc. 40th International Conference on Machine Learning Vol. 202 (eds. Krause, A. et al.) 24950–24962 (PMLR, 2023).
  7. Yang, X., Cheng, W., Petzold, L., Wang, W. Y. & Chen, H. DNA-GPT: divergent N-gram analysis for training-free detection of GPT-generated text. In The Twelfth International Conference on Learning Representations https://openreview.net/forum?id=Xlayxj2fWp(2024).
  8. Bao, G., Zhao, Y., Teng, Z., Yang, L. & Zhang, Y. Fast-DetectGPT: efficient zero-shot detection of machine-generated text via conditional probability curvature. In The Twelfth International Conference on Learning Representations https://openreview.net/forum?id=Bpcgcr8E8Z (ICLR, 2024).
  9. Yu, X. et al. GPT Paternity Test: GPT generated text detection with GPT genetic inheritance. Preprint at https://ar5iv.labs.arxiv.org/html/2305.12519 (2023).
  10. Sadasivan, V. S., Kumar, A., Balasubramanian, S., Wang, W. & Feizi, S. Can AI-generated text be reliably detected? In The Twelfth International Conference on Learning Representations https://openreview.net/forum?id=OOgsAZdFOt (2023).
  11. Jurafsky, D. & Martin, J. H. Speech and Language Processing: An Introduction to Natural Language Processing, Computational Linguistics, and Speech Recognition with Language Models 3rd edn (Prentice Hall, 2025).
  12. arXiv.org submitters. arxiv dataset. kaggle https://www.kaggle.com/datasets/Cornell-University/arxiv/versions/165 (2024).
  13. Fraser, N. rbiorxiv: Client for the ‘bioRxiv’ API. GitHub https://github.com/nicholasmfraser/rbiorxiv (2024).
  14. Nature portfolio. Springer Nature https://www.nature.com/nature-portfolio (2025).
  15. Liang, W. Mapping the increasing use of LLMs in scientific papers. COLM https://openreview.net/forum?id=YX7QnhxESU (2024).

Cited by

See all 82

Metadata

Authors
Liang, W., Zhang, Y., Wu, Z., Lepp, H., Ji, W., Zhao, X., Cao, H., Liu, S., He, S., Huang, Z., Yang, D., Potts, C., Manning, C. D., & Zou, J.
Affiliations
  • Stanford University (US)
  • University of California, Santa Barbara (US)
  • Emory University (US)
Corresponding author
  • Liang, Weixin
  • Zou, James
Venue
Nature Human Behaviour
Published
2025-08-04
Volume
9
Issue
12
Pages
2599-2609
ISSN
2397-3374 (ISSN-L)
OpenAlex
W4412929247
Cited by
80
References
38
Citations by year
2026: 70 · 2025: 10
FWCI
50.74
Type
journal article
Language
en
Open access
closed
Licence
Locations
  • Nature Human Behaviour (publishedVersion)
  • PubMed (publishedVersion)
Topics
  • Social Sciences › Library and Information Sciences › scientometrics and bibliometrics research
  • Social Sciences › Library and Information Sciences › Academic Writing and Publishing
  • Social Sciences › Library and Information Sciences › Academic Publishing and Open Access
Keywords
large language models, LLM-generated text, scientific writing, scholarly publishing, preprints, author productivity, paper length, computer science literature
MeSH terms
Large Language Models, Humans, Language, Publishing, Writing, Bibliometrics

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: Quantifying large language model usage in scientific papers. https://app.cochat.ai/papers/10.1038/s41562-025-02273-8
The paper
Liang, W., Zhang, Y., Wu, Z., Lepp, H., Ji, W., Zhao, X., Cao, H., Liu, S., He, S., Huang, Z., Yang, D., Potts, C., Manning, C. D., & Zou, J. (2025). Quantifying large language model usage in scientific papers. Nature Human Behaviour. https://doi.org/10.1038/s41562-025-02273-8

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.