MétaCan
Menu
Back to cohort
Record W4410521503 · doi:10.1038/s41557-025-01815-x

A framework for evaluating the chemical knowledge and reasoning abilities of large language models against the expertise of chemists

2025· article· en· W4410521503 on OpenAlexfundno aff
A.H. Mirza, Nawaf Alampara, Sreekanth Kunchapu, Martiño Ríos-García, Benedict Emoekabu, Aswanth Krishnan, Mara Schilling-Wilhelmi, Macjonathan Okereke, Anagha Aneesh, Mehrdad Asgari, J. Eberhardt, Amir Mohammad Elahi, Hani M. Elbeheiry, M.V. Gil, Christina Glaubitz, Maximilian Greiner, Caroline T. Holick, Tim Hoffmann, Lea C. Klepsch, Yannik Köster, Fabian Alexander Kreth, Jakob Meyer, Santiago Miret, Jan Matthias Peschel, Michael Ringleb, Nicole C. Roesner, J. Schreiber, Ulrich S. Schubert, Leanne M. Stafast, A. D. Dinga Wonanke, Michael Pieler, Philippe Schwaller, Kevin Maik Jablonka

Bibliographic record

VenueNature Chemistry · 2025
Typearticle
Languageen
FieldMaterials Science
TopicMachine Learning in Materials Science
Canadian institutionsnot available
FundersConsejo Superior de Investigaciones CientíficasHORIZON EUROPE Framework ProgrammeSlezská Univerzita v OpavěHelmholtz AssociationOffice of Multicultural Interests Department of Local Government and CommunitiesMinisterio de Economía y CompetitividadSoutheastern Ontario Academic Medical OrganizationSchweizerischer Nationalfonds zur Förderung der Wissenschaftlichen ForschungEuropean CommissionMinisterio de Ciencia, Innovación y UniversidadesDeutsche ForschungsgemeinschaftUK Research and InnovationAgencia Estatal de InvestigaciónU.S. Department of StateUS-UK Fulbright CommissionFriedrich-Schiller-Universität JenaCarl-Zeiss-StiftungGovernment of the United KingdomFulbright AssociationNational Science Foundation
KeywordsBenchmarkingHarmProcess (computing)Value (mathematics)ChemistryManagement scienceComputer scienceCognitive sciencePsychologyEngineeringSocial psychologyMachine learningProgramming languageManagement

Abstract

fetched live from OpenAlex

Large language models (LLMs) have gained widespread interest owing to their ability to process human language and perform tasks on which they have not been explicitly trained. However, we possess only a limited systematic understanding of the chemical capabilities of LLMs, which would be required to improve models and mitigate potential harm. Here we introduce ChemBench, an automated framework for evaluating the chemical knowledge and reasoning abilities of state-of-the-art LLMs against the expertise of chemists. We curated more than 2,700 question-answer pairs, evaluated leading open- and closed-source LLMs and found that the best models, on average, outperformed the best human chemists in our study. However, the models struggle with some basic tasks and provide overconfident predictions. These findings reveal LLMs' impressive chemical capabilities while emphasizing the need for further research to improve their safety and usefulness. They also suggest adapting chemistry education and show the value of benchmarking frameworks for evaluating LLMs in specific domains.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.

How this classification was reachedexpand

Full frame machine prediction

Teacher imitation

Not calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.

metaresearch head score (Codex)0.015
metaresearch head score (Gemma)0.070
Version: metacan-v3-hybrid-931329e0061cValidation status: machine_predicted_unvalidated
Candidate categoriesMetaresearch
Consensus categoriesnone
DomainCandidate signal: Evaluation · Consensus signal: none
Study designCandidate signal: Simulation or modeling · Consensus signal: Simulation or modeling
GenreCandidate signal: Empirical · Consensus signal: none
Teacher disagreement score0.985
Threshold uncertainty score0.081

Distilled classifier scores by category (both heads)

CategoryCodexGemma
Metaresearch0.0150.070
Meta-epidemiology (narrow)0.0030.001
Meta-epidemiology (broad)0.0010.002
Bibliometrics0.0080.003
Science and technology studies0.0010.002
Scholarly communication0.0030.006
Open science0.0030.004
Research integrity0.0030.003
Insufficient payload (model declined to judge)0.0040.002

Machine scores (provisional)

The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.

Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.

Opus teacher head0.011
GPT teacher head0.357
Teacher spread0.346 · how far apart the two teachers sit on this one work
Validation statusscore_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it

Classification

machine, unvalidated

Machine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.

Study designSimulation or modeling
DomainEvaluation
GenreEmpirical

How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".

Quick stats

Citations55
Published2025
Admission routes1
Has abstractyes

Explore more

Same venueNature ChemistrySame topicMachine Learning in Materials ScienceFrench-language works237,207