Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

  • Correspondence
  • Published:

The dangers of using proprietary LLMs for research

Nature Machine Intelligence volume 6, pages 4–5 (2024)Cite this article

Subjects

The release of ChatGPT at the end of 2022 thrust large language models (LLMs) into the limelight. By enabling its users to query the model directly in natural language, ChatGPT democratized access to these models — a welcome development. Since then, ChatGPT and similar tools such as Bard, Claude and Bing AI have shown their versatility and efficiency on a wide variety of tasks.

Social scientists have been quick to embrace these models. They have used these LLM-based AI assistants to summarize research articles, to debug code, and even to emulate survey participants, experimental subjects or agents in computer simulations1. Researchers have also adopted them to annotate texts. By passing a simple prompt to the machine, they can now sort through thousands of documents. They can do so rapidly, precisely, and according to their own coding scheme2. The promise is, indeed, alluring.

This is a preview of subscription content, access via your institution

Relevant articles

Open Access articles citing this article.

Access options

Buy this article

  • Purchase on Springer Link
  • Instant access to full article PDF
Buy now

Prices may be subject to local taxes which are calculated during checkout

References

  1. Ziems, C. et al. Preprint at https://doi.org/10.48550/arXiv.2305.03514 (2023).

  2. Gilardi, F., Alizadeh, M. & Kubli, M. Proc. Natl Acad. Sci. USA 120, e2305016120 (2023). (2023).

    Article  Google Scholar 

  3. Do, S., Ollion, É. & Shen, R. Sociol. Methods Res. https://doi.org/10.1177/00491241221134526 (2022).

  4. Ollion, É., Shen, R., Macanovic, A. & Chatelain, A. Preprint at SocArXiv https://doi.org/10.31235/osf.io/x58kn (2023).

  5. Reiss, M. V. Preprint at https://doi.org/10.48550/arXiv.2304.11085 (2023).

  6. Rytting, C. M. et al. Preprint at https://doi.org/10.48550/arXiv.2306.02177 (2023).

  7. Spirling, A. Nature 616, 413 (2023).

    Article  Google Scholar 

  8. Lai, V. D. et al. Preprint at https://doi.org/10.48550/arXiv.2304.05613 (2023).

  9. Saphra, N., Fleisig, E., Cho, K. & Lopez, A. Preprint at https://doi.org/10.48550/arXiv.2311.05020 (2023).

  10. Bail, C. A. Preprint at SocArXiv https://doi.org/10.31235/osf.io/rwtzs (2023).

Download references

Author information

Authors and Affiliations

  1. Centre de Recherche en Économie et de Statistiques (CREST), CNRS, École polytechnique, GENES, ENSAE Paris, Institut Polytechnique de Paris, Palaiseau, France

    Étienne Ollion, Rubing Shen & Arnault Chatelain

  2. Médialab, Sciences Po, Paris, France

    Rubing Shen

  3. Department of Sociology, Utrecht University/ICS, Utrecht, The Netherlands

    Ana Macanovic

  4. Centre for Complex Systems Studies, Utrecht University, Utrecht, The Netherlands

    Ana Macanovic

Authors
  1. Étienne Ollion
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Rubing Shen
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Ana Macanovic
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Arnault Chatelain
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Étienne Ollion.

Ethics declarations

Competing interests

The authors declare no competing interests.

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Ollion, É., Shen, R., Macanovic, A. et al. The dangers of using proprietary LLMs for research. Nat Mach Intell 6, 4–5 (2024). https://doi.org/10.1038/s42256-023-00783-6

Download citation

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1038/s42256-023-00783-6

This article is cited by

Search

Advanced search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing