{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,9]],"date-time":"2026-09-09T17:12:56Z","timestamp":1788973976708,"version":"build-2803163510"},"reference-count":51,"publisher":"National Academy of Sciences","issue":"37","license":[{"start":{"date-parts":[[2026,9,9]],"date-time":"2026-09-09T00:00:00Z","timestamp":1788912000000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by-nc-nd\/4.0\/"}],"funder":[{"DOI":"10.13039\/501100001711","name":"Schweizerischer Nationalfonds zur F\u00f6rderung der Wissenschaftlichen Forschung","doi-asserted-by":"publisher","award":["228724"],"award-info":[{"award-number":["228724"]}],"id":[{"id":"10.13039\/501100001711","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":["www.pnas.org"],"crossmark-restriction":true},"short-container-title":["Proc. Natl. Acad. Sci. U.S.A."],"published-print":{"date-parts":[[2026,9,15]]},"abstract":"<jats:p>Large Language Models (LLMs) are entering democratic contexts as instruments of governance, where the challenges at hand are ill-structured, marked by ambiguity and contestation. Ill-structured democratic problems demand more than factual precision; they call for intersubjective reasoning: context-sensitive judgments that others can understand and publicly accept. Using the Deliberative Reason Index (DRI), this study evaluates 60 LLMs against human deliberation across nine policy scenarios. Only four models consistently exceed the permutation-based null benchmark for alignment with human patterns of reason-giving. Most models fall short: their reason\u2013preference structures rarely clear this threshold, even though their outputs can still appear coherent and persuasive. Yet outputs can appear reasonable even when this alignment is absent. The observed gap between surface plausibility and deliberative coherence urges caution: deploying LLMs in governance contexts requires prior assessment of their deliberative reasoning capacity, not just their surface outputs.<\/jats:p>","DOI":"10.1073\/pnas.2600126123","type":"journal-article","created":{"date-parts":[[2026,9,9]],"date-time":"2026-09-09T16:26:36Z","timestamp":1788971196000},"update-policy":"https:\/\/doi.org\/10.1073\/pnas.cm10313","source":"Crossref","is-referenced-by-count":0,"title":["Plausible nonsense and deliberative reasoning: Benchmarking LLMs against human judgment"],"prefix":"10.1073","volume":"123","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-9295-6776","authenticated-orcid":false,"given":"Francesco","family":"Veri","sequence":"first","affiliation":[{"id":[{"id":"https:\/\/ror.org\/02crff812","id-type":"ROR","asserted-by":"publisher"}],"name":"Centre for Democracy Studies Aarau, University of Zurich","place":["Aarau, Switzerland"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-7754-9606","authenticated-orcid":false,"given":"Gustavo","family":"Kreia Umbelino","sequence":"additional","affiliation":[{"id":[{"id":"https:\/\/ror.org\/02crff812","id-type":"ROR","asserted-by":"publisher"}],"name":"Centre for Democracy Studies Aarau, University of Zurich","place":["Aarau, Switzerland"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"341","published-online":{"date-parts":[[2026,9,9]]},"reference":[{"key":"e_1_3_4_1_2","unstructured":"J. Lindsey On the biology of a large language model. Transformer Circuits (2025). https:\/\/transformer-circuits.pub\/2025\/attribution-graphs\/biology.html."},{"key":"e_1_3_4_2_2","doi-asserted-by":"publisher","DOI":"10.1016\/0004-3702(73)90011-8"},{"key":"e_1_3_4_3_2","doi-asserted-by":"publisher","DOI":"10.1017\/9781316344392"},{"key":"e_1_3_4_4_2","doi-asserted-by":"publisher","DOI":"10.1016\/j.tics.2024.04.010"},{"key":"e_1_3_4_5_2","doi-asserted-by":"publisher","DOI":"10.1017\/CBO9780511840005"},{"key":"e_1_3_4_6_2","volume-title":"The Enigma of Reason","author":"Mercier H.","year":"2017","unstructured":"H. Mercier, D. Sperber, The Enigma of Reason (Harvard University Press, 2017)."},{"key":"e_1_3_4_7_2","doi-asserted-by":"publisher","DOI":"10.1111\/j.1467-9221.2012.00873.x"},{"key":"e_1_3_4_8_2","doi-asserted-by":"publisher","DOI":"10.1126\/science.aaw2694"},{"key":"e_1_3_4_9_2","doi-asserted-by":"publisher","DOI":"10.1093\/oso\/9780198820291.001.0001"},{"key":"e_1_3_4_10_2","doi-asserted-by":"publisher","DOI":"10.1017\/S0003055423000023"},{"key":"e_1_3_4_11_2","doi-asserted-by":"publisher","DOI":"10.1177\/14789299241239909"},{"key":"e_1_3_4_12_2","doi-asserted-by":"publisher","DOI":"10.1007\/s44163-024-00111-w"},{"key":"e_1_3_4_13_2","doi-asserted-by":"publisher","DOI":"10.16997\/jdd.1556"},{"key":"e_1_3_4_14_2","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2024.emnlp-main.306"},{"key":"e_1_3_4_15_2","unstructured":"D. Jarrett Language agents as digital representatives in collective decision-making. arXiv [Preprint] (2025). https:\/\/arxiv.org\/abs\/2502.09369 [Accessed 12 July 2025]."},{"key":"e_1_3_4_16_2","volume-title":"AI Poses \u201cRisk of Extinction\u201d Industry Leaders Warn","author":"Roose K.","year":"2023","unstructured":"K. Roose, AI Poses \u201cRisk of Extinction\u201d Industry Leaders Warn (The New York Times, 2023)."},{"key":"e_1_3_4_17_2","doi-asserted-by":"publisher","DOI":"10.1073\/pnas.2518443122"},{"key":"e_1_3_4_18_2","doi-asserted-by":"publisher","DOI":"10.1016\/j.patter.2024.100988"},{"key":"e_1_3_4_19_2","doi-asserted-by":"publisher","DOI":"10.1037\/0033-2909.108.3.480"},{"key":"e_1_3_4_20_2","doi-asserted-by":"publisher","DOI":"10.1007\/s43681-022-00239-4"},{"key":"e_1_3_4_21_2","doi-asserted-by":"publisher","DOI":"10.1093\/oso\/9780192848925.003.0007"},{"key":"e_1_3_4_22_2","doi-asserted-by":"publisher","DOI":"10.1017\/bpp.2024.37"},{"key":"e_1_3_4_23_2","volume-title":"Thinking, Fast and Slow","author":"Kahneman D.","year":"2011","unstructured":"D. Kahneman, Thinking, Fast and Slow (Farrar, Straus and Giroux, ed. 1, 2011).","edition":"1"},{"key":"e_1_3_4_24_2","doi-asserted-by":"publisher","DOI":"10.1086\/296363"},{"key":"e_1_3_4_25_2","first-page":"920","article-title":"Research on solving ill-structured problems for e-learning: Cognitive perspectives","volume":"5","author":"Chen J.","year":"2015","unstructured":"J. Chen, X. Li, Research on solving ill-structured problems for e-learning: Cognitive perspectives. Int. J. Inf. Educ. Technol. 5, 920\u2013923 (2015).","journal-title":"Int. J. Inf. Educ. Technol."},{"key":"e_1_3_4_26_2","first-page":"10","article-title":"All problems are not equal: Implications for problem-based learning","volume":"2","author":"Jonassen D. H.","year":"2008","unstructured":"D. H. Jonassen, W. Hung, All problems are not equal: Implications for problem-based learning. Interdiscip. J. Probl. Based Learn. 2, 10\u201328 (2008).","journal-title":"Interdiscip. J. Probl. Based Learn."},{"key":"e_1_3_4_27_2","doi-asserted-by":"publisher","DOI":"10.1017\/CBO9780511509612"},{"key":"e_1_3_4_28_2","doi-asserted-by":"publisher","DOI":"10.1215\/9780822381815"},{"key":"e_1_3_4_29_2","volume-title":"Genome Editing: Formulating An Australian Community Response: Report to Decision Makers, Stakeholders, and Members of The Public","author":"Nicol D.","year":"2022","unstructured":"D. Nicol , Genome Editing: Formulating An Australian Community Response: Report to Decision Makers, Stakeholders, and Members of The Public (Centre for Law and Genetics, University of Tasmania, 2022)."},{"key":"e_1_3_4_30_2","doi-asserted-by":"publisher","DOI":"10.1561\/105.00000092"},{"key":"e_1_3_4_31_2","doi-asserted-by":"publisher","DOI":"10.1145\/3571730"},{"key":"e_1_3_4_32_2","doi-asserted-by":"crossref","unstructured":"M. Turpin J. Michael E. Perez S. R. Bowman \u201cLanguage models don\u2019t always say what they think: Unfaithful explanations in chain-of-thought prompting\u201d in Advances in Neural Information Processing Systems 74952\u201374965 (2023).","DOI":"10.52202\/075280-3275"},{"key":"e_1_3_4_33_2","first-page":"1","article-title":"ChatGPT is bullshit","volume":"26","author":"Hicks M. T.","year":"2024","unstructured":"M. T. Hicks, J. Humphries, J. Slater, ChatGPT is bullshit. Ethics Inf. Technol. 26, 1\u201310 (2024).","journal-title":"Ethics Inf. Technol."},{"key":"e_1_3_4_34_2","unstructured":"A. Trevisan H. Giddens S. Dillon A. F. Blackwell The BS-meter: A ChatGPT-trained instrument to detect sloppy language-games. arXiv [Preprint] (2024) https:\/\/arxiv.org\/abs\/2411.15129 [Accessed 12 July 2025]."},{"key":"e_1_3_4_35_2","unstructured":"L. Eliot Lawyers are being tripped up by AI sycophancy when using AI to devise legal strategies. Forbes (2026). https:\/\/www.forbes.com\/sites\/lanceeliot\/2026\/04\/04\/lawyers-are-being-tripped-up-by-ai-sycophancy-when-using-ai-to-devise-legal-strategies\/."},{"key":"e_1_3_4_36_2","doi-asserted-by":"publisher","DOI":"10.1126\/science.aec8352"},{"key":"e_1_3_4_37_2","doi-asserted-by":"crossref","unstructured":"B. Jiang \u201cTowards rationality in language and multimodal agents: A survey\u201d in Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers) (Association for Computational Linguistics 2025) pp. 3656\u20133675.","DOI":"10.18653\/v1\/2025.naacl-long.186"},{"key":"e_1_3_4_38_2","unstructured":"J. Gu A survey on LLM-as-a-Judge. arXiv [Preprint] (2024). https:\/\/arxiv.org\/abs\/2411.15594 [Accessed 12 July 2025]."},{"key":"e_1_3_4_39_2","unstructured":"J. Li A survey on benchmarks of multimodal large language models. arXiv [Preprint] (2024). https:\/\/arxiv.org\/abs\/2408.08632 [Accessed 12 July 2025]."},{"key":"e_1_3_4_40_2","unstructured":"M. Zan Y. Zhang B. Zhang F. Liu D. Cheng Can the capability of large language models be described by human ability? A meta study. arXiv [Preprint] (2025). https:\/\/arxiv.org\/abs\/2504.12332 [Accessed 12 July 2025]."},{"key":"e_1_3_4_41_2","doi-asserted-by":"publisher","DOI":"10.1017\/S1755773925000050"},{"key":"e_1_3_4_42_2","doi-asserted-by":"publisher","DOI":"10.1017\/S1755773909990245"},{"key":"e_1_3_4_43_2","doi-asserted-by":"publisher","DOI":"10.1093\/oxfordhb\/9780198747369.013.26"},{"key":"e_1_3_4_44_2","volume-title":"Making It Explicit: Reasoning, Representing and Discursive Commitment","author":"Brandom R. B.","year":"1994","unstructured":"R. B. Brandom, Making It Explicit: Reasoning, Representing and Discursive Commitment (Harvard University Press, 1994)."},{"key":"e_1_3_4_45_2","doi-asserted-by":"publisher","DOI":"10.4159\/9780674028739"},{"key":"e_1_3_4_46_2","doi-asserted-by":"publisher","DOI":"10.7551\/mitpress\/1564.001.0001"},{"key":"e_1_3_4_47_2","doi-asserted-by":"publisher","DOI":"10.1145\/3715070.3749265"},{"key":"e_1_3_4_48_2","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2024.emnlp-main.16"},{"key":"e_1_3_4_49_2","unstructured":"J. Haase Within-model vs between-prompt variability in large language models for creative tasks. arXiv [Preprint] (2026). https:\/\/arxiv.org\/abs\/2601.21339 [Accessed 24 April 2026]."},{"key":"e_1_3_4_50_2","unstructured":"G. K. Umbelino Generating LLM data. GitHub. https:\/\/github.com\/gumbelino\/drillm. Accessed 26 August 2026."},{"key":"e_1_3_4_51_2","unstructured":"S. Niemeyer F. Veri J. S. Dryzek A. B\u00e4chtiger Replication Data for: How Deliberation Happens: Enabling Deliberative Reason. Harvard Dataverse. https:\/\/doi.org\/10.7910\/DVN\/6OKNOY. Accessed 26 August 2026."}],"container-title":["Proceedings of the National Academy of Sciences"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/pnas.org\/doi\/pdf\/10.1073\/pnas.2600126123","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,9,9]],"date-time":"2026-09-09T16:27:05Z","timestamp":1788971225000},"score":1,"resource":{"primary":{"URL":"https:\/\/pnas.org\/doi\/10.1073\/pnas.2600126123"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,9,9]]},"references-count":51,"journal-issue":{"issue":"37","published-print":{"date-parts":[[2026,9,15]]}},"alternative-id":["10.1073\/pnas.2600126123"],"URL":"https:\/\/doi.org\/10.1073\/pnas.2600126123","relation":{},"ISSN":["0027-8424","1091-6490"],"issn-type":[{"value":"0027-8424","type":"print"},{"value":"1091-6490","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,9,9]]},"assertion":[{"value":"2026-01-03","order":0,"name":"received","label":"Received","group":{"name":"publication_history","label":"Publication History"}},{"value":"2026-07-10","order":2,"name":"accepted","label":"Accepted","group":{"name":"publication_history","label":"Publication History"}},{"value":"2026-09-09","order":3,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}],"article-number":"e2600126123"}}