CoChat

A review of large language models and autonomous agents in chemistry

Open access

The DOI matches its Crossref record; 623 of the 623 references it cites were found in a registry or an index.

Abstract

This review examines the roles of large language models (LLMs) and autonomous agents in chemistry, exploring advancements in molecule design, property prediction, and synthesis automation.

Overview

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1039/d4sc03921a

Cites

623 of 623 references verified0 citation errors · 0 not checkable

Statistics

Figures

Fig. 1 AI-powered LLMs accelerate chemical discovery with models that address key challenges in property prediction, property directed molecule generation, and synthesis prediction. Autonomous agents connect these models and additional tools thereby enabling rapid exploration of vast chemical spaces.
Fig. 1 AI-powered LLMs accelerate chemical discovery with models that address key challenges in property prediction, property directed molecule generation, and synthesis prediction. Autonomous agents connect these models and additional tools thereby enabling rapid exploration of vast chemical spaces.CC BY · PubMed Central
Fig. 2 (a) The generalized encoder–decoder transformer: the encoder on the left converts an input into a vector, while the decoder on the right predicts the next token in a sequence. (b) Encoder–decoder transformers are traditionally used for translation tasks and, in chemistry, for reaction prediction, translating reactants into products. (c) Encoder-only transformers provide a vector output and are typically used for sentiment analysis. In chemistry, they are used for property prediction or classification tasks. (d) Decoder-only transformers generate likely next tokens in a sequence. In chemistry, they are used to generate new molecules given an instruction and description of molecules.
Fig. 2 (a) The generalized encoder–decoder transformer: the encoder on the left converts an input into a vector, while the decoder on the right predicts the next token in a sequence. (b) Encoder–decoder transformers are traditionally used for translation tasks and, in chemistry, for reaction prediction, translating reactants into products. (c) Encoder-only transformers provide a vector output and are typically used for sentiment analysis. In chemistry, they are used for property prediction or classification tasks. (d) Decoder-only transformers generate likely next tokens in a sequence. In chemistry, they are used to generate new molecules given an instruction and description of molecules.CC BY · PubMed Central
Fig. 3 Classification of LLMs in chemistry and biochemistry according to their application.
Fig. 3 Classification of LLMs in chemistry and biochemistry according to their application.CC BY · PubMed Central
Fig. 4 Illustration of how Large Language Models (LLMs) evolved chronologically. The dates display the first publication of each model.
Fig. 4 Illustration of how Large Language Models (LLMs) evolved chronologically. The dates display the first publication of each model.CC BY · PubMed Central
Fig. 5 Number of training tokens (on log scale) available from various chemical sources compared with typical LLM training runs. The numbers are drawn from ZINC,165 PubChem,166 Touvron et al.,167 ChEMBL,168 and Kinney et al.169
Fig. 5 Number of training tokens (on log scale) available from various chemical sources compared with typical LLM training runs. The numbers are drawn from ZINC,165 PubChem,166 Touvron et al.,167 ChEMBL,168 and Kinney et al.169CC BY · PubMed Central
Fig. 6 Agent's architecture as defined in this review. According to our definition, an agent is composed of a central program (typically an LLM and the code to implement the agent's dynamic behavior) and the agent modules. The agent continuously receives observations from the environment and decides which action should be executed to complete the task given to it. Here, we define the agent as the set of elements whose decision is trainable, that is, the LLM, the agent code, the decision process, and the agent modules. Given a task, the agent uses the agent modules (memory, reasoning, planning, profiling) and the LLM to decide which action should be executed. This action is executed by calling a tool from the environment. After the action is executed, an observation is produced and fed back to the agent. The agent can use perception to receive inputs in different modalities from the environment. (A) Description of agent modules, (B) illustration of the agent architecture, (C) illustration of the environment components, (D) description of tools elements present in the environment.
Fig. 6 Agent's architecture as defined in this review. According to our definition, an agent is composed of a central program (typically an LLM and the code to implement the agent's dynamic behavior) and the agent modules. The agent continuously receives observations from the environment and decides which action should be executed to complete the task given to it. Here, we define the agent as the set of elements whose decision is trainable, that is, the LLM, the agent code, the decision process, and the agent modules. Given a task, the agent uses the agent modules (memory, reasoning, planning, profiling) and the LLM to decide which action should be executed. This action is executed by calling a tool from the environment. After the action is executed, an observation is produced and fed back to the agent. The agent can use perception to receive inputs in different modalities from the environment. (A) Description of agent modules, (B) illustration of the agent architecture, (C) illustration of the environment components, (D) description of tools elements present in the environment.CC BY · PubMed Central

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

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?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Research works printed but not linked: 31

  1. Willett, 2011, Wiley Interdiscip. Rev.: Comput. Mol. Sci., 1, 46
  2. C.Hansch and A.Leo , Exploring QSAR.: Fundamentals and applications in chemistry and biology , American Chemical Society , 1995 , vol. 1
  3. Segler, 2018, ACS Cent. Sci., 4, 120, 10.1021/acscentsci.7b00512 · DOI · Paper page
  4. Prakash Dwivedi, 2023, J. Mach. Learn. Res., 24, 1
  5. Sanchez-Lengeling, 2021, Distill, 6, e33
  6. J.Gilmer , S. S.Schoenholz , P. F.Riley , O.Vinyals and E. D.George , Neural message passing for quantum chemistry , in International conference on machine learning , PMLR , 2017 , pp. 1263–1272
  7. V.Nair and G. E.Hinton , Rectified linear units improve restricted boltzmann machines , in Proceedings of the 27th international conference on machine learning , ICML-10 , 2010 , pp. 807–814
  8. Stiennon, 2020, Adv. Neural Inf. Process. Syst., 33, 3008
  9. T.Hayes , R.Rao , H.Akin , N. J.Sofroniew , D.Oktay , Z.Lin , R.Verkuil , V. Q.Tran , J.Deaton , M.Wiggert , R.Badkundri , I.Shafkat , J.Gong , A.Derry , R. S.Molina , N.Thomas , Y.Khan , C.Mishra , C.Kim , L. J.Bartie , M.Nemeth , P. D.Hsu , T.Sercu , S.Candido and A.Rives , bioRxiv , 2024 , preprint, 10.1101/2024.07.01.600583v2 , https://www.biorxiv.org/content/10.1101/2024.07.01.600583v2 · doi:10.1101/2024.07.01.600583v2
  10. Li, 2021, 2021, Proc. Int. Wirel. Commun. Mob. Comput. Conf., 1530, 10.1155/2021/7181815 · DOI
  11. A.Radford , K.Narasimhan , T.Salimans and I.Sutskever , Improving language understanding with unsupervised learning , 2018
  12. Hu, 2020, Adv. Neural Inf. Process. Syst., 33, 22118
  13. N.O'Boyle and A.Dalke , DeepSMILES: An Adaptation of SMILES for Use in Machine-Learning of Chemical Structures , ChemRxiv , 2018 , preprint
  14. D.Hendrycks , C.Burns , S.Basart , A.Zou , M.Mazeika , D.Song and J.Steinhardt , Measuring massive multitask language understanding , in International Conference on Learning Representations , 2020 , https://openreview.net/pdf?id=d7KBjmI3GmQ
  15. A.Pal , L.Kumar Umapathi and M.Sankarasubbu , MedMCQA: A large-scale multi-subject multi-choice dataset for medical domain question answering , in Conference on Health, Inference and Learning , PMLR , 2022 , pp. 248–260
  16. S.Wang , Y.Guo , Y.Wang , H.Sun and J.Huang , SMILES-BERT: Large Scale Unsupervised Pre-Training for Molecular Property Prediction , in Proceedings of the 10th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics, BCB ’19 , Association for Computing Machinery , New York, NY, USA , 2019 , pp. 429–436 , 10.1145/3307339.3342186 · DOI · Paper page
  17. Ross, 2022, Nat. Mach. Intell., 4, 1256, 10.1038/s42256-022-00580-7 · DOI · Paper page
  18. J.Vig , Bertviz: A tool for visualizing multihead self-attention in the bert model , in ICLR workshop: Debugging machine learning models , 2019 , vol. 3
  19. Rong, 2020, Adv. Neural Inf. Process. Syst., 33, 12559
  20. M.Xu , X.Yuan , S.Miret and J.Tang , ProtST: Multi-modality learning of protein sequences and biomedical texts , in International Conference on Machine Learning , PMLR , 2023 , pp. 38749–38767
  21. C.-Y.Lin , ROUGE: A Package for Automatic Evaluation of Summaries , in Text Summarization Branches Out , Association for Computational Linguistics , Barcelona, Spain , 2004 , pp. 74–81 , https://aclanthology.org/W04-1013
  22. David, 2020, J. Cheminform., 12, 56, 10.1186/s13321-020-00460-5 · DOI · Paper page
  23. Lo, 2018, Drug Discovery Today, 23, 1538, 10.1016/j.drudis.2018.05.010 · DOI · Paper page
  24. Y.Liu , M.Ott , N.Goyal , J.Du , M.Joshi , D.Chen , O.Levy , M.Lewis , L.Zettlemoyer and V.Stoyanov , RoBERTa: A robustly optimized BERT pretraining approach , arXiv , 2019 , preprint, arXiv:1907.11692, 10.48550/arXiv.1907.11692 , http://arxiv.org/abs/1907.11692 · DOI · Paper page
  25. B. S.Bloom , Taxonomy of Educational Objectives: The Classification of Educational Goals; Handbook. Cognitive Domain , McKay , 1968
  26. S. S.Raman , V.Cohen , E.Rosen and I.Idrees , Planning with large language models via corrective re-prompting , Foundation Models , 2022 , https://openreview.net/pdf?id=cMDMRBe1TKs
  27. B.Ranković and P.Schwaller , BoChemian: Large language model embeddings for Bayesian optimization of chemical reactions , 2023 , https://openreview.net/pdf?id=A1RVn1m3J3
  28. Zheng, 2023, J. Am. Chem. Soc., 145, 18048, 10.1021/jacs.3c05819 · DOI · Paper page
  29. G.Landrum , Rdkit documentation, Release, 1(1-79):4 , 2013 , pp. 15333752 , https://media.readthedocs.org/pdf/rdkit/latest/rdkit.pdf
  30. Schwaller, 2021, Sci. Adv., 7, eabe4166, 10.1126/sciadv.abe4166 · DOI · Paper page
  31. O.Schilter , M.Alberts , F.Zipoli , A. C.Vaucher , P.Schwaller and T.Laino , Unveiling the secrets of 1H-NMR spectroscopy: A novel approach utilizing attention mechanisms , 2023 , https://openreview.net/pdf?id=4ilKwquW51

Cited by

See all 361

Metadata

Authors
Ramos, M. C., Collison, C. J., & White, A. D.
Affiliations
  • Future House, Inc. (US)
  • Rochester Institute of Technology (US)
Corresponding author
White, Andrew D.
Venue
Chemical Science
Published
2025
Volume
16
Issue
6
Pages
2514-2572
ISSN
  • 2041-6520 (ISSN-L)
  • 2041-6539
OpenAlex
W4405185373
Cited by
274
References
516
Citations by year
2026: 162 · 2025: 106
FWCI
6.54
Type
journal article
Language
en
Open access
Locations
  • Chemical Science (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • arXiv (Cornell University) (submittedVersion, cc-by)
  • arXiv (Cornell University) (cc-by)
Funders
  • Basic Energy Sciences (DE-SC0023354)
  • Rochester Institute of Technology
Topics
  • Physical Sciences › Materials Science › Materials Chemistry › Machine Learning in Materials Science
  • Life Sciences › Pharmacology, Toxicology and Pharmaceutics › Drug Discovery › Computational Drug Discovery Methods
  • Physical Sciences › Computer Science › Artificial Intelligence › Artificial Intelligence Applications
Keywords
large language models, autonomous agents, chemistry, property prediction, synthesis planning, laboratory automation, model interpretability, LLM benchmark

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

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.