How should the advancement of large language models affect the practice of science?
Proceedings of the National Academy of Sciences · 04.02.2025 · 122 (5)
DOI: 10.1073/pnas.2401227121 ↗Kurz gefasst
Fachartikel in der Zeitschrift Proceedings of the National Academy of Sciences, online erschienen am 27.01.2025. Verfasst von Marcel Binz und 17 weiteren Autor:innen. Beteiligte Einrichtung: Max Planck Institute for Biological Cybernetics, Tübingen, Baden-Württemberg 72076, Germany. Hauptthemen laut MEDLINE: Large Language Models, Science, Diffusion of Innovation. Die Arbeit wurde bisher 85-mal zitiert (Crossref) und nennt 89 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Large language models (LLMs) are being increasingly incorporated into scientific workflows. However, we have yet to fully grasp the implications of this integration. How should the advancement of large language models affect the practice of science? For this opinion piece, we have invited four diverse groups of scientists to reflect on this query, sharing their perspectives and engaging in debate. Schulz et al. make the argument that working with LLMs is not fundamentally different from working with human collaborators, while Bender et al. argue that LLMs are often misused and overhyped, and that their limitations warrant a focus on more specialized, easily interpretable tools. Marelli et al. emphasize the importance of transparent attribution and responsible use of LLMs. Finally, Botvinick and Gershman advocate that humans should retain responsibility for determining the scientific roadmap. To facilitate the discussion, the four perspectives are complemented with a response from each group. By putting these different perspectives in conversation, we aim to bring attention to important considerations within the academic community regarding the adoption of LLMs and their impact on both current and future scientific practices.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Max Planck Institute for Biological Cybernetics, Tübingen, Baden-Württemberg 72076, Germany.
- Helmholtz Center for Computational Health, Munich, Oberschleißheim, Bayern 85764, Germany.
- Department of Computer Science, Technical University of Munich, München, Bayern 80333, Germany.
- Munich Center for Machine Learning, München, Bayern 80333, Germany.
- Department of Psychological and Brain Sciences, University of California, Santa Barbara, CA 93106.
- Institute of Psychology, Eötvös Loránd University, Budapest 1053, Hungary.
ORCID-Profile: Marcel Binz, Stephan Alaniz, Balazs Aczel, Carl T Bergstrom, Colin Allen, Daniel Schad, Jevin D West, Richard M Shiffrin, Samuel J Gershman, Vencislav Popov, Emily M Bender, Marco Marelli, Zeynep Akata
Zitieren
APA
Binz, M., Alaniz, S., Roskies, A., Aczel, B., Bergstrom, C. T., Allen, C., Schad, D., Wulff, D., West, J. D., Zhang, Q., Shiffrin, R. M., Gershman, S. J., Popov, V., Bender, E. M., Marelli, M., Botvinick, M. M., Akata, Z., & Schulz, E. (2025). How should the advancement of large language models affect the practice of science?. Proceedings of the National Academy of Sciences, 122(5). https://doi.org/10.1073/pnas.2401227121
Vancouver
Binz M, Alaniz S, Roskies A, Aczel B, Bergstrom CT, Allen C, et al. How should the advancement of large language models affect the practice of science?. Proceedings of the National Academy of Sciences. 2025;122(5). doi:10.1073/pnas.2401227121
BibTeX
@article{Binz2025,
author = {Binz, Marcel and Alaniz, Stephan and Roskies, Adina and Aczel, Balazs and Bergstrom, Carl T and Allen, Colin and Schad, Daniel and Wulff, Dirk and West, Jevin D and Zhang, Qiong and Shiffrin, Richard M and Gershman, Samuel J and Popov, Vencislav and Bender, Emily M and Marelli, Marco and Botvinick, Matthew M and Akata, Zeynep and Schulz, Eric},
title = {How should the advancement of large language models affect the practice of science?},
journal = {Proceedings of the National Academy of Sciences},
year = {2025},
volume = {122},
number = {5},
doi = {10.1073/pnas.2401227121}
}
EUVIMED