Pith. sign in

REVIEW 4 major objections 5 minor 45 references

Generative Artificial Intelligence Extracts Structure-Function Relationships from Plants for New Materials

T0 review · 4 major / 5 minor · reviewed 2026-08-05 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read The paper claims that a generative-AI pipeline can extract structure-property relationships from plant and biomimetics literature and translate them into a new, laboratory-validated pollen-based adhesive.

desk verdict Plausible end-to-end AI-to-material loop, but the abstract leaves outcome leakage and missing numbers unresolved; worth a serious referee to check the protocol. read the letter →

arxiv 2508.06591 v1 pith:45XPEMA2 submitted 2025-08-08 cs.LG cond-mat.dis-nncond-mat.mtrl-scicond-mat.othercs.AIcs.CL

Rachel K. Luu , Jingyu Deng , Mohammed Shahrudin Ibrahim , Nam-Joon Cho , Ming Dao , Subra Suresh , Markus J. Buehler This is my paper
classification cs.LGcond-mat.dis-nncond-mat.mtrl-scicond-mat.othercs.AIcs.CL
keywords generativeAIlargelanguagemodelsbioinspiredmaterialsretrieval-augmentedgenerationstructure-propertyrelationshipspollen-basedadhesiveplantsciencediscovery
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper tries to establish that a generative-AI pipeline, not a hand-built expert system, can read across plant science, biomimetics, and materials engineering to propose new materials that actually work in the lab. It combines a fine-tuned language model, retrieval-augmented generation, agentic workflows, and a hierarchical sampling scheme to generate and screen hundreds of hypotheses about humidity-responsive structures such as pollen and lady palm leaves. The proof is a laboratory result: the AI-generated procedure, design, and mechanical prediction led to a novel pollen-based adhesive with tunable morphology and measured shear strength. A sympathetic reader would see the contribution as closing the loop from literature to fabrication.

What carries the argument

The load-bearing mechanism is the integrated AI pipeline: BioinspiredLLM is a language model fine-tuned on bioinspired design literature; retrieval-augmented generation grounds each response in retrieved documents; agentic systems refine hypotheses through structured inference protocols; and hierarchical sampling starts broad and narrows to experimentally tractable candidates. The defining move is that a single user query yields hundreds of explicitly reasoned hypotheses, each with a design, procedure, and predicted mechanical property, so the laboratory can act as the final filter.

What would settle it

Run the pipeline on a held-out humidity-responsive plant system and execute the highest-ranked protocol without human edits. The central claim fails if the protocol cannot be followed as written, if the predicted shear strength falls outside the measured error bars, or if the winning pollen recipe appears verbatim in pre-existing plant-adhesive papers that the retrieval step simply copied.

Watch

Extended reading notes

Core claim

The paper's central claim is that a generative AI system can make a materially useful scientific discovery by transferring structure-property relationships across fields. Concretely, the authors assert that their fine-tuned language model (BioinspiredLLM), aided by retrieval-augmented generation, agentic reasoning, and a hierarchical sampling strategy, produced procedures, materials designs, and mechanical predictions for humidity-responsive materials. Those outputs were then implemented in a laboratory and resulted in a pollen-based adhesive with tunable morphology and measured shear strength. If correct, this shows that AI-assisted ideation can go beyond producing plausible text and can gu

Load-bearing premise

The paper's load-bearing premise is that the LLM-generated recipes and mechanical predictions are faithful to the plant-science literature and were chosen before the lab results were known, rather than retrofitted.

Editorial extensions

If this is right

  • If the claim holds, AI systems can generate experimental procedures and property predictions that survive wet-lab validation, not just plausible text.
  • The framework supplies a route from one query to hundreds of ranked, tractable hypotheses, changing the economics of bioinspired materials ideation.
  • Cross-domain transfer from plant structures to synthetic adhesives becomes a routine retrieval-plus-reasoning task rather than requiring a single domain expert.
  • The pollen-based adhesive with tunable morphology and measured shear strength is itself evidence that plant-derived adhesives can be designed through this workflow.
  • Human-AI collaboration is centered on selecting among machine-generated, lab-testable designs rather than on generating designs from scratch.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A direct extension the paper leaves implicit: the same retrieval-and-ranking loop could be pointed at other humidity-responsive or mechanically adaptive natural systems, with the lab validation step acting as the final filter.
  • A testable ablation, not reported in the paper, would remove RAG, the agentic refinement, or the hierarchical sampler one at a time and measure how many candidate hypotheses remain experimentally reproducible; this would reveal which component actually carries the discovery.
  • The shear-strength number invites a comparative baseline: whether a human literature review restricted to the same sources would propose an equivalent adhesive; if so, the pipeline's value would be speed and coverage rather than novelty.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The paper claims to present an end-to-end generative-AI framework for materials discovery, combining a fine-tuned LLM (BioinspiredLLM), retrieval-augmented generation, agentic systems, and a hierarchical sampling strategy. The stated application is the extraction of structure-property relationships from plant science and biomimetics literature and their translation into experimentally validated materials. The abstract reports that LLM-generated procedures, designs, and mechanical predictions were tested in the laboratory, culminating in a novel pollen-based adhesive with tunable morphology and measured shear strength. The provided full text is not legible in the submitted encoding, and no quantitative experimental values, controls, or selection protocol appear in the abstract.

Significance. If the central claim is substantiated, this would be a noteworthy demonstration of an AI-driven loop from disparate literature to a fabricated material, with practical relevance to bioinspired adhesive design. The paper's intended contribution is timely and potentially interesting. However, the current submission does not provide enough evidence to assess the claim: there are no numerical shear-strength values, replicate counts, error bars, control adhesives, or a demonstrated pre-experimental selection rule. The significance is therefore conditional on the authors supplying the missing experimental and procedural evidence in a readable form.

major comments (4)
  1. [Abstract (validation claim)] The abstract states that structured inference protocols 'generate and evaluate hundreds of hypotheses' and that the pipeline 'culminat[ed] in' a pollen-based adhesive. This wording does not establish that the adhesive was selected and its properties predicted before laboratory results were known. If the reported winner was chosen after the experiments, the 'hundreds' of hypotheses become a multiple-comparison selection problem and the AI-discovery claim collapses into post-hoc rationalization. The authors must report the temporal order, the frozen candidate list, the ranking rule, how many candidates were actually tested, and the pass/fail rate across all tested hypotheses.
  2. [Abstract (experimental evidence)] The abstract asserts 'measured shear strength' but gives no value, unit, sample size, standard deviation, or comparison to a control adhesive. This is the central experimental evidence for the 'validated through real-world implementation' claim. The authors need to report the actual shear strength, the number of replicates, the test geometry, and a comparison against a non-bioinspired control or a reference material. Without these numbers, the validation claim is uncheckable.
  3. [Full text (integrity and readability)] The supplied full text carries the header 'arXiv:2508.06594v1 [econ.GN]' rather than the manuscript's stated ID (arXiv:2508.06591, cs.LG), and the body text is largely undecodable. No equation, data table, or experimental section can be independently audited. This is not merely a formatting issue: it prevents verification of the central experimental claims. The authors should resubmit a readable manuscript with the correct identifier and a traceable data/appendix section.
  4. [RAG/LLM faithfulness] The paper claims to 'extract' structure-property relationships from the literature, but the abstract provides no evidence that the retrieved sources actually support the causal mechanisms used in the successful adhesive design. LLM-generated text can be plausible without being causally faithful to the cited literature. The authors should provide citation-level traces for the key structure-property relationships, an explicit expert-verification step, and a statement of how hallucinated mechanisms were excluded before laboratory testing.
minor comments (5)
  1. [Methods] Specify the base model and version of BioinspiredLLM, the fine-tuning dataset, and the RAG corpus size and retrieval parameters. Without these details, reproducibility is impossible.
  2. [Hierarchical Sampling] Define the 'Hierarchical Sampling strategy' more precisely and explain what distinguishes it from standard search/ranking over LLM outputs. A small illustrative example or pseudo-code would help.
  3. [Materials characterization] Provide units and uncertainties for all reported material properties, including adhesion strength, morphology parameters, and humidity response. State how many independent samples were measured and by which instrument.
  4. [Related work and novelty] The phrase 'first-of-its-kind' needs support from a systematic comparison with prior LLM-for-materials-discovery work. Please add a dedicated related-work section and position the contribution relative to existing AI-driven ideation and bioinspired materials pipelines.
  5. [Reproducibility artifacts] Make the prompts, retrieval traces, generated hypotheses, and experimental protocols available in a supplementary repository or appendix so that reviewers and readers can check the pipeline's behavior end to end.

Circularity Check

0 steps flagged · score 0.0 of 10

No identifiable circularity: the central claim rests on external laboratory validation, not on a self-referential derivation or fitted prediction.

full rationale

The paper's claimed chain is empirical rather than derivational: plant/biomimetics literature -> LLM/RAG hypotheses -> laboratory testing -> a pollen-based adhesive. The only quoted support for the central claim is that 'LLM-generated procedures, materials designs, and mechanical predictions were tested in the laboratory, culminating in the fabrication of a novel pollen-based adhesive with tunable morphology and measured shear strength.' That is an external benchmark (lab measurement), not an internal equation. No step in the readable material defines a predicted quantity in terms of the measured quantity, fits a parameter to the same data it later 'predicts,' or appeals to a self-citation or uniqueness theorem. The encoded/garbled full text cannot be parsed into equations, so no specific Equation X = Equation Y reduction can be exhibited. The residual concern about hypothesis selection after seeing lab results ('outcome leakage') is a temporal/statistical validity issue, not circularity of definition or self-citation; without evidence of such leakage, it cannot be scored as circularity under the hard rules. Honest finding: no significant circularity.

Assumptions & free parameters 0 free parameters · 3 assumptions · 0 invented entities

No equations or fitted values are visible in the abstract, so no free parameters can be audited; the full text likely contains hyperparameters and sampling thresholds, but they are inaccessible in the garbled copy. No new physical entities are posited: BioinspiredLLM is a software artifact, not an invented physical entity. The axioms listed are the unstated domain assumptions that the abstract's claims depend on.

assumptions (3)
  • domain assumption Plant science literature contains structure-function relationships that are textually encoded and can be correctly extracted by an LLM with RAG.
    The entire pipeline presupposes faithful retrieval and interpretation of the plant and biomimetics corpus. If the LLM hallucinates or the retrieval misses key papers, the extracted relationships are unreliable. Invoked implicitly in the abstract's claim that the framework extracts structure-property relationships from the literature.
  • domain assumption The biomimetic transfer from plant structures to a synthetic pollen-based adhesive preserves the causal mechanism, e.g., humidity-driven morphology change affecting adhesion.
    The validation step only means the adhesive works; the paper's framing requires that it works for the reason the AI extracted. Without mechanistic verification, the structure-function link is correlational.
  • domain assumption The hypothesis-evaluation protocol ranks generated ideas by experimental tractability and promise without using knowledge of the eventual lab outcome.
    The abstract says hundreds of hypotheses were generated and evaluated and that the work culminated in one successful adhesive. If the selection rule leaked outcome knowledge, the demonstration is circular. No selection rule is described in the abstract.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Generative Artificial Intelligence Extracts Structure-Function Relationships from Plants for New Materials." pith.science (2026). https://pith.science/paper/45XPEMA2

@misc{pith2026250806591,
  author       = {Pith},
  title        = {Pith review of: Generative Artificial Intelligence Extracts Structure-Function Relationships from Plants for New Materials},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/45XPEMA2}},
  note         = {Machine review of arXiv:2508.06591}
}
read the original abstract

Large language models (LLMs) have reshaped the research landscape by enabling new approaches to knowledge retrieval and creative ideation. Yet their application in discipline-specific experimental science, particularly in highly multi-disciplinary domains like materials science, remains limited. We present a first-of-its-kind framework that integrates generative AI with literature from hitherto-unconnected fields such as plant science, biomimetics, and materials engineering to extract insights and design experiments for materials. We focus on humidity-responsive systems such as pollen-based materials and Rhapis excelsa (broadleaf lady palm) leaves, which exhibit self-actuation and adaptive performance. Using a suite of AI tools, including a fine-tuned model (BioinspiredLLM), Retrieval-Augmented Generation (RAG), agentic systems, and a Hierarchical Sampling strategy, we extract structure-property relationships and translate them into new classes of bioinspired materials. Structured inference protocols generate and evaluate hundreds of hypotheses from a single query, surfacing novel and experimentally tractable ideas. We validate our approach through real-world implementation: LLM-generated procedures, materials designs, and mechanical predictions were tested in the laboratory, culminating in the fabrication of a novel pollen-based adhesive with tunable morphology and measured shear strength, establishing a foundation for future plant-derived adhesive design. This work demonstrates how AI-assisted ideation can drive real-world materials design and enable effective human-AI collaboration.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

45 extracted references · 31 canonical work pages

  1. [1]

    , author Luu, R

    author Lu, W. , author Luu, R. K. & author Buehler, M. J. title Fine-tuning large language models for domain adaptation: exploration of training strategies, scaling, model merging and synergistic capabilities . journal npj Computational Materials volume 11 ( year 2025 ). ://www.nature.com/articles/s41524-025-01130-2. note Accessed 2025-08-07

    work page 2025

  2. [2]

    author Chen, J. et al. title Can LLMs' Tuning Methods Work in Medical Multimodal Domain? pages 112--122 ( year 2024 ). ://arxiv.org/abs/2403.06407

    arXiv 2024

  3. [3]

    author Van Herck, J. et al. title Assessment of fine-tuned large language models for real-world chemistry and material science applications . journal Chemical Science volume 16 , pages 670--684 ( year 2025 ). ://pubs.rsc.org/en/content/articlehtml/2025/sc/d4sc04401k https://pubs.rsc.org/en/content/articlelanding/2025/sc/d4sc04401k

    work page 2025

  4. [4]

    author Rozi \` e re, B. et al. title Code Llama: Open Foundation Models for Code ( year 2023 ). ://arxiv.org/abs/2308.12950v3

    arXiv 2023

  5. [5]

    author Luu, R. K. & author Buehler, M. J. title BioinspiredLLM: Conversational Large Language Model for the Mechanics of Biological and Bio-Inspired Materials . journal Advanced Science pages 2306724 ( year 2023 )

    work page 2023

  6. [6]

    & author Burgert, I

    author Speck, T. & author Burgert, I. title Plant Stems: Functional Design and Mechanics . journal Annual Review of Materials Research volume 41 , pages 169--193 ( year 2011 ). ://doi.org/10.1146/annurev-matsci-062910-100425

    work page doi:10.1146/annurev-matsci-062910-100425 2011

  7. [7]

    author Gibson, L. J. title The hierarchical structure and mechanics of plant materials . journal Journal of The Royal Society Interface volume 9 , pages 2749--2766 ( year 2012 ). ://royalsocietypublishing.org/doi/full/10.1098/rsif.2012.0341

    work page arXiv 2012

  8. [8]

    , author Mail, M

    author Barthlott, W. , author Mail, M. , author Bhushan, B. & author Koch, K. title Plant Surfaces: Structures and Functions for Biomimetic Innovations . journal Nano-Micro Letters 2016 9:2 volume 9 , pages 1--40 ( year 2017 ). ://link-springer-com.libproxy.mit.edu/article/10.1007/s40820-016-0125-1

    work page doi:10.1007/s40820-016-0125-1 2016

Show all 45 references
  1. [9]

    author Zhao, Z. L. , author Zhou, S. , author Feng, X. Q. & author Xie, Y. M. title On the internal architecture of emergent plants . journal Journal of the Mechanics and Physics of Solids volume 119 , pages 224--239 ( year 2018 )

    2018

  2. [10]

    author Bar-On, B. et al. title Structural origins of morphing in plant tissues . journal Applied Physics Letters volume 105 , pages 33703 ( year 2014 ). /aip/apl/article/105/3/033703/931948/Structural-origins-of-morphing-in-plant-tissues

    2014

  3. [11]

    & author Fratzl, P

    author Burgert, I. & author Fratzl, P. title Actuation systems in plants as prototypes for bioinspired devices . journal Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences volume 367 , pages 1541--1557 ( year 2009 ). ://royalsoci...

    2009

  4. [12]

    author Sadeghi, A. et al. title A plant-inspired robot with soft differential bending capabilities . journal Bioinspiration & Biomimetics volume 12 , pages 015001 ( year 2016 ). ://iopscience.iop.org/article/10.1088/1748-3190/12/1/015001 https://iopscience.iop.org/article/10.1...

    2016 doi

  5. [13]

    title In touch: plant responses to mechanical stimuli

    author Braam, J. title In touch: plant responses to mechanical stimuli . journal New Phytologist volume 165 , pages 373--389 ( year 2005 ). ://onlinelibrary.wiley.com/doi/full/10.1111/j.1469-8137.2004.01263.x https://onlinelibrary.wiley.com/doi/abs/10.1111/j.1469-8137.2004.012...

    2005

  6. [14]

    title The light eaters : the new science of plants ( year 2024 )

    author Schlanger, Z. title The light eaters : the new science of plants ( year 2024 ). ://books.google.com/books/about/The_Light_Eaters_The_New_Science_of_Plan.html?id=PMXbEAAAQBAJ

    2024

  7. [15]

    author Zhao, Z. et al. title Actuation and locomotion driven by moisture in paper made with natural pollen . journal Proceedings of the National Academy of Sciences of the United States of America volume 117 , pages 8711--8718 ( year 2020 ). ://www.pnas.org/doi/abs/10.1073/pna...

    2020 doi

  8. [16]

    author Zhao, Z. et al. title Digital printing of shape-morphing natural materials . journal Proceedings of the National Academy of Sciences of the United States of America volume 118 , pages e2113715118 ( year 2021 ). ://www.pnas.org/doi/abs/10.1073/pnas.2113715118

    2021 doi

  9. [17]

    author Deng, J. et al. title Plant-Based Shape Memory Cryogel for Hemorrhage Control . journal Advanced Materials volume 36 , pages 2311684 ( year 2024 ). ://onlinelibrary.wiley.com/doi/full/10.1002/adma.202311684 https://onlinelibrary.wiley.com/doi/abs/10.1002/adma.202311684 ...

    2024 doi

  10. [18]

    , author Liu, M

    author Guo, K. , author Liu, M. , author Vella, D. , author Suresh, S. & author Hsia, K. J. title Dehydration-induced corrugated folding in Rhapis excelsa plant leaves . journal Proceedings of the National Academy of Sciences of the United States of America volume 121 , pages ...

    2024 doi

  11. [19]

    author Zhang, F. et al. title Unperceivable motion mimicking hygroscopic geometric reshaping of pine cones . journal Nature Materials volume 21 , pages 1357--1365 ( year 2022 ). ://www.nature.com/articles/s41563-022-01391-2

    2022

  12. [20]

    author Luu, R. K. et al. title Learning from Nature to Achieve Material Sustainability: Generative AI for Rigorous Bio-inspired Materials Design . journal An MIT Exploration of Generative AI ( year 2024 ). ://mit-genai.pubpub.org/pub/jpkwte2n/release/2

    2024

  13. [21]

    title A Survey on Hallucination in Large Language Models: Principles, Taxonomy, Challenges, and Open Questions

    author HuangLei et al. title A Survey on Hallucination in Large Language Models: Principles, Taxonomy, Challenges, and Open Questions . journal ACM Transactions on Information Systems volume 43 , pages 1--55 ( year 2025 ). ://dl-acm-org.libproxy.mit.edu/doi/10.1145/3703155

    2025 doi

  14. [22]

    , author Drobnjak, A

    author Perkovi \' c , G. , author Drobnjak, A. & author Boti c ki, I. title Hallucinations in LLMs: Understanding and Addressing Challenges . journal 2024 47th ICT and Electronics Convention, MIPRO 2024 - Proceedings pages 2084--2088 ( year 2024 )

    2024

  15. [23]

    author Buehler, M. J. title P R ef L ex O R : Preference-based recursive language modeling for exploratory optimization of reasoning and agentic thinking . journal npj Artificial Intelligence ( year 2024 ). ://arxiv.org/abs/2410.12375

    2024 arXiv

  16. [24]

    author Buehler, M. J. title In-situ graph reasoning and knowledge expansion using G raph- P R ef L ex O R . journal Advanced Intelligent Discovery ( year 2025 ). ://arxiv.org/abs/2501.08120

    2025 arXiv

  17. [25]

    author Buehler, M. J. title Graph-aware isomorphic attention for adaptive dynamics in transformers . journal APL Machine Learning ( year 2025 ). ://arxiv.org/abs/2501.02393

    2025 arXiv

  18. [26]

    author Lewis, P. et al. title Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks . journal Advances in Neural Information Processing Systems volume 2020-December ( year 2020 ). ://arxiv.org/abs/2005.11401v4

    2020 arXiv

  19. [27]

    author Buehler, M. J. title Generative Retrieval-Augmented Ontologic Graph and Multiagent Strategies for Interpretive Large Language Model-Based Materials Design . journal ACS Engineering Au volume 4 , pages 241--277 ( year 2024 ). ://pubs.acs.org/doi/full/10.1021/acsengineeri...

    2024 doi

  20. [28]

    & author Buehler, M

    author Ni, B. & author Buehler, M. J. title MechAgents: Large language model multi-agent collaborations can solve mechanics problems, generate new data, and integrate knowledge ( year 2023 ). ://arxiv.org/abs/2311.08166v1

    2023 arXiv

  21. [29]

    & author Buehler, M

    author Ghafarollahi, A. & author Buehler, M. J. title Rapid and Automated Alloy Design with Graph Neural Network-Powered LLM-Driven Multi-Agent Systems ( year 2024 ). ://arxiv.org/abs/2410.13768v1

    2024 arXiv

  22. [30]

    & author Buehler, M

    author Ghafarollahi, A. & author Buehler, M. J. title ProtAgents: protein discovery via large language model multi-agent collaborations combining physics and machine learning . journal Digital Discovery volume 3 , pages 1389--1409 ( year 2024 ). ://pubs.rsc.org/en/content/arti...

    2024

  23. [31]

    , author Buehler, M

    author Ghafarollahi, A. , author Buehler, M. J. , author Ghafarollahi, A. & author Buehler, M. J. title SciAgents: Automating Scientific Discovery Through Bioinspired Multi-Agent Intelligent Graph Reasoning . journal Advanced Materials pages 2413523 ( year 2024 ). ://onlinelib...

    2024 doi

  24. [32]

    & author Buehler, M

    author Ghafarollahi, A. & author Buehler, M. J. title AtomAgents: Alloy design and discovery through physics-aware multi-modal multi-agent artificial intelligence ( year 2024 ). ://arxiv.org/abs/2407.10022v1

    2024 arXiv

  25. [33]

    & author Williams, P

    author G \' o mez-Rodr \' i guez, C. & author Williams, P. title A Confederacy of Models: a Comprehensive Evaluation of LLMs on Creative Writing . journal Findings of the Association for Computational Linguistics: EMNLP 2023 pages 14504--14528 ( year 2023 ). ://arxiv.org/abs/2...

    2023 arXiv

  26. [34]

    author Zhao, Y. et al. title Assessing and Understanding Creativity in Large Language Models ( year 2024 ). ://arxiv.org/abs/2401.12491v1

    2024 arXiv

  27. [35]

    author Boden, M. A. title The Creative Mind: Myths and Mechanisms ( publisher Routledge , year 2004 ). ://books-google-com.libproxy.mit.edu/books?id=KOR_AgAAQBAJ&lr=&source=gbs_navlinks_s

    2004

  28. [36]

    author Guilford, J. P. title The structure of intellect . journal Psychological Bulletin volume 53 , pages 267--293 ( year 1956 )

    1956

  29. [37]

    author Touvron, H. et al. title Llama 2: Open Foundation and Fine-Tuned Chat Models

  30. [38]

    author Naleway, S. E. , author Porter, M. M. , author McKittrick, J. & author Meyers, M. A. title Structural Design Elements in Biological Materials: Application to Bioinspiration . journal Advanced Materials volume 27 , pages 5455--5476 ( year 2015 ). ://onlinelibrary.wiley.c...

    2015 doi

  31. [39]

    , author McKittrick, J

    author Chen, P.-Y. , author McKittrick, J. & author Meyers, M. A. title Biological materials: Functional adaptations and bioinspired designs . journal Progress in Materials Science volume 57 , pages 1492--1704 ( year 2012 ). ://linkinghub.elsevier.com/retrieve/pii/S0079642512000242

    2012

  32. [40]

    author Meyers, M. A. , author McKittrick, J. & author Chen, P.-Y. title Structural Biological Materials: Critical Mechanics-Materials Connections . journal Science volume 339 , pages 773--779 ( year 2013 ). ://www.sciencemag.org/lookup/doi/10.1126/science.1220854

    2013 doi

  33. [41]

    author Achiam, J. et al. title Gpt-4 technical report . journal arXiv preprint arXiv:2303.08774 ( year 2023 )

    2023 arXiv

  34. [42]

    author Liang, P. et al. title Holistic evaluation of language models . journal arXiv preprint arXiv:2211.09110 ( year 2022 )

    2022 arXiv

  35. [43]

    & author Lee, H.-y

    author Chiang, C.-H. & author Lee, H.-y. title Can large language models be an alternative to human evaluations? journal arXiv preprint arXiv:2305.01937 ( year 2023 )

    2023 arXiv

  36. [44]

    author Zhang, Y. et al. title Llmeval: A preliminary study on how to evaluate large language models . In booktitle Proceedings of the AAAI Conference on Artificial Intelligence , vol. volume 38 , pages 19615--19622 ( year 2024 )

    2024

  37. [45]

    , author Chern, E

    author Chern, S. , author Chern, E. , author Neubig, G. & author Liu, P. title Can large language models be trusted for evaluation? scalable meta-evaluation of llms as evaluators via agent debate . journal arXiv preprint arXiv:2401.16788 ( year 2024 )

    2024 arXiv

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.