AI can help humans find common ground in democratic deliberation
The DOI matches its Crossref record; 25 of the 25 references it cites were found in a registry or an index.
Abstract
Finding agreement through a free exchange of views is often difficult. Collective deliberation can be slow, difficult to scale, and unequally attentive to different voices. In this study, we trained an artificial intelligence (AI) to mediate human deliberation. Using participants’ personal opinions and critiques, the AI mediator iteratively generates and refines statements that express common ground among the group on social or political issues. Participants ( N = 5734) preferred AI-generated statements to those written by human mediators, rating them as more informative, clear, and unbiased. Discussants often updated their views after the deliberation, converging on a shared perspective. Text embeddings revealed that successful group statements incorporated dissenting voices while respecting the majority position. These findings were replicated in a virtual citizens’ assembly involving a demographically representative sample of the UK population.
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.
Integrity signals
Registry
Cites
Statistics
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.
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
BLEU: a method for automatic evaluation of machine translation
Papineni, K., Roukos, S., Ward, T., & Zhu, W.-J. (2001) · Proceedings of the 40th Annual Meeting on Association for Computational Linguistics - ACL '02 · doi:10.3115/1073083.1073135 · Cited by 21,344
Decoupled Weight Decay Regularization
Loshchilov, I., & Hutter, F. (2017) · arXiv · doi:10.48550/arxiv.1711.05101 · Cited by 12,522
Mixed-Effects Models in Sand S-PLUS
Pinheiro, J. C., & Bates, D. M. (2000) · Statistics and Computing · doi:10.1007/978-1-4419-0318-1 · Cited by 11,187
Random effects structure for confirmatory hypothesis testing: Keep it maximal
Barr, D. J., Levy, R., Scheepers, C., & Tily, H. J. (2013) · Journal of Memory and Language · doi:10.1016/j.jml.2012.11.001 · Cited by 10,328 · cited 1×
“Inferential statistics were based on generalized linear mixed-effects models, implemented in R ( 46, 47 ), adopting the maximal random-effects structure that our data permits in each case ( 48 ).”
MethodTraining language models to follow instructions with human feedback
Ouyang, L., Wu, J., Jiang, X., Almeida, D., Wainwright, C. L., Mishkin, P., Zhang, C., Agarwal, S., Slama, K., Ray, A., et al. (2022) · arXiv · doi:10.48550/arxiv.2203.02155 · Cited by 4,208
Algorithm aversion: People erroneously avoid algorithms after seeing them err.
Dietvorst, B. J., Simmons, J. P., & Massey, C. (2015) · Journal of Experimental Psychology: General · doi:10.1037/xge0000033 · Cited by 3,232 · cited 1×
“Moreover, there is room for healthy ,andsomepeoplemay haveaprincipledaversiontotheideathat politicalorsocialideascanbegeneratedbya computer( 40, 41 ), which may temper enthusi-asm for our approach.”
BackgroundSIMR : an R package for power analysis of generalized linear mixed models by simulation
Green, P., & MacLeod, C. J. (2016) · Methods in Ecology and Evolution · doi:10.1111/2041-210x.12504 · Cited by 2,467
Rank Analysis of Incomplete Block Designs: I. The Method of Paired Comparisons
Bradley, R. A., & Terry, M. E. (1952) · Biometrika · doi:10.2307/2334029 · Cited by 2,412
The robust beauty of improper linear models in decision making.
Dawes, R. M. (1979) · American Psychologist · doi:10.1037/0003-066x.34.7.571 · Cited by 1,966
Universal Sentence Encoder for English
Cer, D., Yang, Y., Kong, S.-Y., Hua, N., Limtiaco, N., St. John, R., Constant, N., Guajardo-Cespedes, M., Yuan, S., Tar, C., et al. (2018) · Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing: System Demonstrations · doi:10.18653/v1/d18-2029 · Cited by 1,750
RANK ANALYSIS OF INCOMPLETE BLOCK DESIGNS: THE METHOD OF PAIRED COMPARISONS
BRADLEY, R. A., & TERRY, M. E. (1952) · Biometrika · doi:10.1093/biomet/39.3-4.324 · Cited by 1,434
Ideology, motivated reasoning, and cognitive reflection
Kahan, D. M. (2013) · Judgment and Decision Making · doi:10.1017/s1930297500005271 · Cited by 1,302 · cited 2×
“The HMthusoffersanewapproachtocollective deliberationthatcircumventssomeofthelim-itationsofin-persondeliberation,includingits cost,limitedscale,thepotentialformediator bias,andpronenesstosocialdesirabilityeffects orinequalityofcontribution( 6 , 7 ).”
BackgroundThe effect of severity of initiation on liking for a group.
Aronson, E., & Mills, J. (1959) · The Journal of Abnormal and Social Psychology · doi:10.1037/h0047195 · Cited by 1,297 · cited 1×
“One risk in this evaluation is that revised statements might be preferred simply because they were encountered later and after a slight additional investment of effort, driving an effort justification bias ( 27 ).”
BackgroundStalnaker, R. (2002) · Linguistics and Philosophy · doi:10.1023/a:1020867916902 · Cited by 1,293
Deep reinforcement learning from human preferences
Christiano, P., Leike, J., Brown, T. B., Martic, M., Legg, S., & Amodei, D. (2017) · arXiv · doi:10.48550/arxiv.1706.03741 · Cited by 1,058
Mixed-Effects Models in S and S-PLUS
(2000) · Statistics and Computing · doi:10.1007/b98882 · Cited by 930
Training Compute-Optimal Large Language Models
Hoffmann, J., Borgeaud, S., Mensch, A., Buchatskaya, E., Cai, T., Rutherford, E., Casas, D. D. L., Hendricks, L. A., Welbl, J., Clark, A., et al. (2022) · arXiv · doi:10.48550/arxiv.2203.15556 · Cited by 929
LaMDA: Language Models for Dialog Applications
Thoppilan, R., De Freitas, D., Hall, J., Shazeer, N., Kulshreshtha, A., Cheng, H.-T., Jin, A., Bos, T., Baker, L., Du, Y., et al. (2022) · arXiv · doi:10.48550/arxiv.2201.08239 · Cited by 778 · cited 2×
“A generative model that has undergone supervised fine-tuning proposes high-quality candidates for the “ group statement, ” and a personalized reward model (PRM) scores the candidates according to the predicted preferences of each individual group member ( 22 – 24 ) (Fig.”
BackgroundMethodPlackett, R. L. (1975) · Applied Statistics · doi:10.2307/2346567 · Cited by 731
Ordinal Regression Models in Psychology: A Tutorial
Bürkner, P.-C., & Vuorre, M. (2019) · Advances in Methods and Practices in Psychological Science · doi:10.1177/2515245918823199 · Cited by 689 · cited 1×
“We fit ordinal models for analyses relating the Likert measurements ( 49 ).”
MethodSentEval: An Evaluation Toolkit for Universal Sentence Representations
Conneau, A., & Kiela, D. (2018) · arXiv · doi:10.48550/arxiv.1803.05449 · Cited by 477
Feng, S., Park, C. Y., Liu, Y., & Tsvetkov, Y. (2023) · Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · doi:10.18653/v1/2023.acl-long.656 · Cited by 149
Argyle, L. P., Bail, C. A., Busby, E. C., Gubler, J. R., Howe, T., Rytting, C., Sorensen, T., & Wingate, D. (2023) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.2311627120 · Cited by 144 · cited 1×
“…10 ), generate statements that showcase the full range of a group ’ s political views ( 11 ), temper partisan discussion by suggestingequanimous rewrites ( 12 ), and mitigate gender imbalance in debate participation by positively contributing to discussions about sensitive issues ( 13 ).”
BackgroundImproving alignment of dialogue agents via targeted human judgements
Glaese, A., McAleese, N., Trębacz, M., Aslanides, J., Firoiu, V., Ewalds, T., Rauh, M., Weidinger, L., Chadwick, M., Thacker, P., et al. (2022) · arXiv · doi:10.48550/arxiv.2209.14375 · Cited by 134
Self-citationA Long Way to Go: Investigating Length Correlations in RLHF
Singhal, P., Goyal, T., Xu, J., & Durrett, G. (2023) · arXiv · doi:10.48550/arxiv.2310.03716 · Cited by 3
Cited by
LLMs-as-Judges: A Comprehensive Survey on LLM-based Evaluation Methods
Li, H., Dong, Q., Chen, J., Su, H., Zhou, Y., Ai, Q., Ye, Z., & Liu, Y. (2024) · arXiv · doi:10.48550/arxiv.2412.05579 · Cited by 40
Toward a science of human–AI teaming for decision making: a complementarity framework
Gonzalez, C., Donahue, K., Goldstein, D. G., Heidari, H., Jalali, M. S., Schelble, B., Singh, A., & Woolley, A. W. (2026) · PNAS Nexus · doi:10.1093/pnasnexus/pgag030 · Cited by 33
“For example, AI monitors that ask targeted clarifying questions during democratic deliberation can help participants identify shared ground on contentious issues (25).”
Increasing happiness through conversations with artificial intelligence
Heffner, J., Qin, C., Chadwick, M., Knutsen, C., Summerfield, C., Kurth-Nelson, Z., & Rutledge, R. B. (2025) · arXiv · doi:10.48550/arxiv.2504.02091 · Cited by 10
Self-citationVergne, J. P. (2026) · Journal of Management Studies · doi:10.1111/joms.70048 · Cited by 1
Metadata
- Authors
- Tessler, M. H., Bakker, M. A., Jarrett, D., Sheahan, H., Chadwick, M. J., Koster, R., Evans, G., Campbell-Gillingham, L., Collins, T., Parkes, D. C., Botvinick, M., & Summerfield, C.
- Venue
- Science
- Published
- 2024-10-18
- Volume
- 386
- Issue
- 6719
- Pages
- eadq2852-eadq2852
- ISSN
- 0036-8075 (ISSN-L)
- 1095-9203
- OpenAlex
- W4403511027
- Cited by
- 195
- References
- 25
- FWCI
- 80.49
- Type
- journal article
- Language
- en
- Open access
- closed
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.