Scientific hypothesis generation by large language models: Laboratory validation in breast cancer treatment
Artikel i vetenskaplig tidskrift, 2025

Large language models (LLMs) have transformed artificial intelligence (AI) and achieved breakthrough performance on a wide range of tasks. In science, the most interesting application of LLMs is for hypothesis formation. A feature of LLMs, which results from their probabilistic structure, is that the output text is not necessarily a valid inference from the training text. These are termed 'hallucinations', and are harmful in many applications. In science, some hallucinations may be useful: novel hypotheses whose validity may be tested by laboratory experiments. Here, we experimentally test the application of LLMs as a source of scientific hypotheses using the domain of breast cancer treatment. We applied the LLM GPT4 to hypothesize novel synergistic pairs of US Food and Drug Administration (FDA)-approved non-cancer drugs that target the MCF7 breast cancer cell line relative to the non-tumorigenic breast cell line MCF10A. In the first round of laboratory experiments, GPT4 succeeded in discovering three drug combinations (out of 12 tested) with synergy scores above the positive controls. GPT4 then generated new combinations based on its initial results, this generated three more combinations with positive synergy scores (out of four tested). We conclude that LLMs are a valuable source of scientific hypotheses.

drug discovery

artificial intelligence for science

machine learning

cancer research

personalized medicine

Författare

Abbi Abdel-Rehim

University of Cambridge

Hector Zenil

Alan Turing Institute

The Francis Crick Institute

University of Oxford

King's College London

University of Cambridge

Oghenejokpeme I. Orhobor

University of Cambridge

Marie Fisher

Arctoris Ltd

Ross J. Collins

Arctoris Ltd

Elizabeth Bourne

Arctoris Ltd

Gareth W. Fearnley

Arctoris Ltd

Emma Tate

Arctoris Ltd

Holly X. Smith

Arctoris Ltd

Larisa N. Soldatova

Goldsmiths, University of London

Ross King

Chalmers, Data- och informationsteknik, Data Science och AI

Göteborgs universitet

University of Cambridge

Journal of the Royal Society Interface

1742-5689 (ISSN) 1742-5662 (eISSN)

Vol. 22 227 20240674

Ämneskategorier (SSIF 2025)

Cancer och onkologi

DOI

10.1098/rsif.2024.0674

Mer information

Senast uppdaterat

2025-06-16