Exploring the role of large language models in the scientific method: from hypothesis to discovery
Abstract We review how Large Language Models (LLMs) are redefining the scientific method and explore their potential applications across different stages of the scientific cycle, from hypothesis testing to discovery. We conclude that, for LLMs to serve as relevant and effective creative engines and productivity enhancers, their deep integration into all steps of the scientific process should be pursued in collaboration and alignment with human scientific goals, with clear evaluation metrics.
Authors
- Alexander Lavin (ORCID: https://orcid.org/0000-0003-3422-7820)
- Sumeer Ahmad Khan
- Héctor Zenil (ORCID: https://orcid.org/0000-0003-0634-4384)
- A. K. M. Firoj Mahmud (ORCID: https://orcid.org/0000-0002-6541-1500)
- James H. Evans
- Jeremy G. Frey (ORCID: https://orcid.org/0000-0003-0842-4302)
- Jesper Tegnér (ORCID: https://orcid.org/0000-0002-9568-5588)
- Alan Bundy (ORCID: https://orcid.org/0000-0002-0578-6474)
- Jared Dunnmon (ORCID: https://orcid.org/0000-0002-4002-0562)
- Sašo Džeroski (ORCID: https://orcid.org/0000-0003-2363-712X)
- Michael Levin (ORCID: https://orcid.org/0000-0001-7292-8084)
- Huck Yang
- Yanbo Zhang (ORCID: https://orcid.org/0000-0002-7627-488X)
Institutions
- Tufts University (US)
- Santa Fe Institute (US)
- Karolinska University Hospital (SE)
- King's College London (GB)
- Zuse Institute Berlin (DE)
- Science for Life Laboratory (SE)
- The Francis Crick Institute (GB)
- Karolinska Institutet (SE)
- Jožef Stefan Institute (SI)
- University of Chicago (US)
- Nvidia (United States) (US)
- The Alan Turing Institute (GB)
- Centre for Sustainable Healthcare (GB)
- University of Southampton (GB)
- King Abdullah University of Science and Technology (SA)
- Network Rail (GB)
- Stanford University (US)
- University of Edinburgh (GB)
Publication Details
- Journal
- npj Artificial Intelligence
- Published
- 2025-08-05
- DOI
- https://doi.org/10.1038/s44387-025-00019-5
- Citations
- 29
- Primary Topic
- Topic Modeling
- Type
- article
- Field-Weighted Citation Impact
- 71.95