What large language models know and what people think they know
Open accessThe DOI matches its Crossref record; 49 of the 49 references it cites were found in a registry or an index.
Abstract
As artificial intelligence systems, particularly large language models (LLMs), become increasingly integrated into decision-making processes, the ability to trust their outputs is crucial. To earn human trust, LLMs must be well calibrated such that they can accurately assess and communicate the likelihood of their predictions being correct. Whereas recent work has focused on LLMs’ internal confidence, less is understood about how effectively they convey uncertainty to users. Here we explore the calibration gap, which refers to the difference between human confidence in LLM-generated answers and the models’ actual confidence, and the discrimination gap, which reflects how well humans and models can distinguish between correct and incorrect answers. Our experiments with multiple-choice and short-answer questions reveal that users tend to overestimate the accuracy of LLM responses when provided with default explanations. Moreover, longer explanations increased user confidence, even when the extra length did not improve answer accuracy. By adjusting LLM explanations to better reflect the models’ internal confidence, both the calibration gap and the discrimination gap narrowed, significantly improving user perception of LLM accuracy. These findings underscore the importance of accurate uncertainty communication and highlight the effect of explanation length in influencing user trust in artificial-intelligence-assisted decision-making environments.
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
Figures








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.
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?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“AI is exceedingly confident and, unlike people, rarely indicates anything less than 100% confidence in the output,they do not use indicators like hedging language for example [48].”
From Incorporating generative AI into a writing-intensive undergraduate course without off-loading learning (2025) · cites this paper as [48]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“However, current VLMs struggle to communicate their internal uncertainty through natural language [69, 117].”
From "It's trained by non-disabled people": Evaluating How Image Quality Affects Product Captioning with Vision-Language Models (2026) · cites this paper as [117]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Previous studies also show that users tend to associate larger models and longer results with greater precision and reliability [25], [26], [27]”
From Hollow-LLM Attack: Computationally Trivial Weights in Zero-Knowledge Verification of LLM Inference (2026) · cites this paper as [25]
Experimental — accuracy not yet measuredIn the citing paper
“and explanations can increase users’ confidence in AI outputs without improving the accuracy of their judgments (Steyvers et al., 2025).”
From Cognitive transparency and trust calibration in explainable AI systems: a psychological model of human-AI decision reliance (2026) · cites this paper as Steyvers et al., 2025
In the citing paper
“This concern is supported by recent evidence that users often overestimate the accuracy of AI outputs, particularly when explanations are provided (Steyvers et al., 2025)”
From Cognitive transparency and trust calibration in explainable AI systems: a psychological model of human-AI decision reliance (2026) · cites this paper as Steyvers et al., 2025
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
Huang, L., Yu, W., Ma, W., Zhong, W., Feng, Z., Wang, H., Chen, Q., Peng, W., Feng, X., Qin, B., et al. (2025) · ACM Transactions on Information Systems · doi:10.1145/3703155 · Cited by 4,688 · cited 1×
“One notable issue is the possibility of generating responses that, while convincing, may be inaccurate or nonsensical 9,10 .”
Training 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
Training Language Models to Follow Instructions with Human Feedback
Ouyang, L., Wu, J., Jiang, X., Almeida, D., Wainwright, C., Mishkin, P., Zhang, C., Agarwal, S., Slama, K., Ray, A., et al. (2022) · Advances in Neural Information Processing Systems 35 · doi:10.52202/068431-2011 · Cited by 3,117
Huang, L., Yu, W., Ma, W., Zhong, W., Feng, Z., Wang, H., Chen, Q., Peng, W., Feng, X., Qin, B., et al. (2023) · arXiv · doi:10.48550/arxiv.2311.05232 · Cited by 2,516 · cited 1×
“One notable issue is the possibility of generating responses that, while convincing, may be inaccurate or nonsensical 9,10 .”
Default Bayes factors for ANOVA designs
Rouder, J. N., Morey, R. D., Speckman, P. L., & Province, J. M. (2012) · Journal of Mathematical Psychology · doi:10.1016/j.jmp.2012.08.001 · Cited by 1,800
Petty, R. E., & Cacioppo, J. T. (1984) · Journal of Personality and Social Psychology · doi:10.1037/0022-3514.46.1.69 · Cited by 1,727 · cited 1×
“This result is consistent with studies in social psychology and communication research that suggest that longer answers or explanations may be perceived as more persuasive or credible, even when they do not contain more meaningful information 27,28 .”
TriviaQA: A Large Scale Distantly Supervised Challenge Dataset for Reading Comprehension
Joshi, M., Choi, E., Weld, D., & Zettlemoyer, L. (2017) · Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · doi:10.18653/v1/p17-1147 · Cited by 1,690
TriviaQA: A Large Scale Distantly Supervised Challenge Dataset for Reading Comprehension
Joshi, M., Choi, E., Weld, D. S., & Zettlemoyer, L. (2017) · arXiv · doi:10.48550/arxiv.1705.03551 · Cited by 1,650
Fleming, S. M., & Lau, H. C. (2014) · Frontiers in Human Neuroscience · doi:10.3389/fnhum.2014.00443 · Cited by 1,408 · cited 1×
“The use of AUC in this context parallels various metrics in psychology for metacognitive discrimination or sensitivity, which similarly aim to evaluate the effectiveness of confidence scores in distinguishing between correct and incorrect answers 47 .”
Buçinca, Z., Malaya, M. B., & Gajos, K. Z. (2021) · arXiv · doi:10.48550/arxiv.2102.09692 · Cited by 1,124
Obtaining well calibrated probabilities using Bayesian binning
Pakdaman Naeini, M., Cooper, G., & Hauskrecht, M. (2015) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v29i1.9602 · Cited by 991
Detecting hallucinations in large language models using semantic entropy
Farquhar, S., Kossen, J., Kuhn, L., & Gal, Y. (2024) · Nature · doi:10.1038/s41586-024-07421-0 · Cited by 934 · cited 4×
“13,17).”
InfluentialTraining 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) · Advances in Neural Information Processing Systems 35 · doi:10.52202/068431-2176 · Cited by 932 · cited 1×
“Overall, there has been little research examining user confidence in LLM output.”
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
What ChatGPT and generative AI mean for science
Stokel-Walker, C., & Van Noorden, R. (2023) · Nature · doi:10.1038/d41586-023-00340-6 · Cited by 828
Does the Whole Exceed its Parts? The Effect of AI Explanations on Complementary Team Performance
Bansal, G., Wu, T., Zhou, J., Fok, R., Nushi, B., Kamar, E., Ribeiro, M. T., & Weld, D. (2021) · Proceedings of the 2021 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3411764.3445717 · Cited by 757 · cited 1×
“Therefore, participants who considered themselves more knowledgeable about a topic were not more adept at estimating the LLM’s performance in that area.”
Does the Whole Exceed its Parts? The Effect of AI Explanations on Complementary Team Performance
Bansal, G., Wu, T., Zhou, J., Fok, R., Nushi, B., Kamar, E., Ribeiro, M. T., & Weld, D. S. (2020) · arXiv · doi:10.48550/arxiv.2006.14779 · Cited by 754
Can LLMs Express Their Uncertainty? An Empirical Evaluation of Confidence Elicitation in LLMs
Xiong, M., Hu, Z., Lu, X., Li, Y., Fu, J., He, J., & Hooi, B. (2023) · arXiv · doi:10.48550/arxiv.2306.13063 · Cited by 618
Beyond the Imitation Game: Quantifying and extrapolating the capabilities of language models
Srivastava, A., Rastogi, A., Rao, A., Shoeb, A. A. M., Abid, A., Fisch, A., Brown, A. R., Santoro, A., Gupta, A., Garriga-Alonso, A., et al. (2022) · arXiv · doi:10.48550/arxiv.2206.04615 · Cited by 569
Using ChatGPT to write patient clinic letters
Ali, S. R., Dobbs, T. D., Hutchings, H. A., & Whitaker, I. S. (2023) · The Lancet Digital Health · doi:10.1016/s2589-7500(23)00048-1 · Cited by 375 · cited 1×
“These models are increasingly integrated into areas such as public health 6 , coding 7 and education 8 .”
Oppenheimer, D. M. (2006) · Applied Cognitive Psychology · doi:10.1002/acp.1178 · Cited by 353
The interpretation of IPCC probabilistic statements around the world
Budescu, D. V., Por, H.-H., Broomell, S. B., & Smithson, M. (2014) · Nature Climate Change · doi:10.1038/nclimate2194 · Cited by 291
Measuring Massive Multitask Language Understanding
Hendrycks, D., Burns, C., Basart, S., Zou, A., Mazeika, M., Song, D., & Steinhardt, J. (2020) · arXiv · doi:10.48550/arxiv.2009.03300 · Cited by 291
Towards Human-Centered Explainable AI: A Survey of User Studies for Model Explanations
Rong, Y., Leemann, T., Nguyen, T.-T., Fiedler, L., Qian, P., Unhelkar, V., Seidel, T., Kasneci, G., & Kasneci, E. (2024) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2023.3331846 · Cited by 278
Scaling Language Models: Methods, Analysis & Insights from Training Gopher
Rae, J. W., Borgeaud, S., Cai, T., Millican, K., Hoffmann, J., Song, F., Aslanides, J., Henderson, S., Ring, R., Young, S., et al. (2021) · arXiv · doi:10.48550/arxiv.2112.11446 · Cited by 235 · cited 1×
“The MMLU dataset is widely used to measure a text model’s multitask accuracy, as it challenges models on their real-world text understanding beyond mere linguistic comprehension, thus making it a robust benchmark for model evaluation 19,41,42 .”
Three Challenges for AI-Assisted Decision-Making
Steyvers, M., & Kumar, A. (2024) · Perspectives on Psychological Science · doi:10.1177/17456916231181102 · Cited by 217
Self-citationRelying on the Unreliable: The Impact of Language Models’ Reluctance to Express Uncertainty
Zhou, K., Hwang, J., Ren, X., & Sap, M. (2024) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2024.acl-long.198 · Cited by 207
Misremembrance of Options Past: Source Monitoring and Choice
Mather, M., Shafir, E., & Johnson, M. K. (2000) · Psychological Science · doi:10.1111/1467-9280.00228 · Cited by 202
Language Models (Mostly) Know What They Know
Kadavath, S., Conerly, T., Askell, A., Henighan, T., Drain, D., Perez, E., Schiefer, N., Hatfield-Dodds, Z., DasSarma, N., Tran-Johnson, E., et al. (2022) · arXiv · doi:10.48550/arxiv.2207.05221 · Cited by 187
Jiang, Z., Araki, J., Ding, H., & Neubig, G. (2021) · Transactions of the Association for Computational Linguistics · doi:10.1162/tacl_a_00407 · Cited by 177 · cited 5×
“This method for reading out model confidence allows a direct computation of the relative probabilities of different possible answers in multiple-choice questions 12,13,18,19,23 and the probability that the answer to an open-ended question is correct 13,17 .”
InfluentialTian, K., Mitchell, E., Zhou, A., Sharma, A., Rafailov, R., Yao, H., Finn, C., & Manning, C. (2023) · Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing · doi:10.18653/v1/2023.emnlp-main.330 · Cited by 123
Smith-Renner, A., Fan, R., Birchfield, M., Wu, T., Boyd-Graber, J., Weld, D. S., & Findlater, L. (2020) · Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3313831.3376624 · Cited by 112
Better Uncertainty Calibration Via Proper Scores for Classification and Beyond
Gruber, S., & Buettner, F. (2022) · neural information processing systems · doi:10.52202/068431-0627 · Cited by 108
What can AI do for me?: evaluating machine learning interpretations in cooperative play
Feng, S., & Boyd-Graber, J. (2019) · Proceedings of the 24th International Conference on Intelligent User Interfaces · doi:10.1145/3301275.3302265 · Cited by 92
Do Large Language Models Know What They Don’t Know?
Yin, Z., Sun, Q., Guo, Q., Wu, J., Qiu, X., & Huang, X. (2023) · Findings of the Association for Computational Linguistics: ACL 2023 · doi:10.18653/v1/2023.findings-acl.551 · Cited by 75
Effects of Explanations in AI-Assisted Decision Making: Principles and Comparisons
Wang, X., & Yin, M. (2022) · ACM Transactions on Interactive Intelligent Systems · doi:10.1145/3519266 · Cited by 74
Improving the communication of uncertainty in climate science and intelligence analysis
Ho, E., Budescu, D. V., Dhami, M. K., & Mandel, D. R. (2015) · Behavioral Science & Policy · doi:10.1353/bsp.2015.0015 · Cited by 73
From nCoder to ChatGPT: From Automated Coding to Refining Human Coding
Zambrano, A. F., Liu, X., Barany, A., Baker, R. S., Kim, J., & Nasiar, N. (2023) · Communications in computer and information science · doi:10.1007/978-3-031-47014-1_32 · Cited by 69
Improving the communication of uncertainty in climate science and intelligence analysis
Ho, E. H., Budescu, D., Dhami, M. K., & Mandel, D. R. (2015) · Behavioral Science & Policy · doi:10.1177/237946151500100206 · Cited by 61
Relying on the Unreliable: The Impact of Language Models' Reluctance to Express Uncertainty
Zhou, K., Hwang, J. D., Ren, X., & Sap, M. (2024) · arXiv · doi:10.48550/arxiv.2401.06730 · Cited by 44
Navigating the Grey Area: How Expressions of Uncertainty and Overconfidence Affect Language Models
Zhou, K., Jurafsky, D., & Hashimoto, T. (2023) · Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing · doi:10.18653/v1/2023.emnlp-main.335 · Cited by 38
Peer reviews of peer reviews: A randomized controlled trial and other experiments
Goldberg, A., Stelmakh, I., Cho, K., Oh, A., Agarwal, A., Belgrave, D., & Shah, N. B. (2025) · PLOS ONE · doi:10.1371/journal.pone.0320444 · Cited by 18 · cited 1×
“In addition, the calibration error was comparable between the two groups (ECE 0.289 versus 0.292).”
InfluentialVerbosity Bias in Preference Labeling by Large Language Models
Saito, K., Wachi, A., Wataoka, K., & Akimoto, Y. (2023) · arXiv · doi:10.48550/arxiv.2310.10076 · Cited by 15 · cited 1×
“Therefore, participants who considered themselves more knowledgeable about a topic were not more adept at estimating the LLM’s performance in that area.”
Peer Reviews of Peer Reviews: A Randomized Controlled Trial and Other ExperimentsPreprint
Goldberg, A., Stelmakh, I., Cho, K., et al. (2023) · arXiv (Cornell University) · doi:10.48550/arxiv.2311.09497 · Cited by 7
How experts and novices judge other people’s knowledgeability from language use
Bower, A. H., Han, N., Soni, A. K., Eckstein, M. P., & Steyvers, M. (2024) · Psychonomic Bulletin & Review · doi:10.3758/s13423-023-02433-9 · Cited by 6
Self-citationTian, K., Mitchell, E., Zhou, A., Sharma, A., Rafailov, R., Yao, H., Finn, C., & Manning, C. D. (2023) · arXiv · doi:10.48550/arxiv.2305.14975 · Cited by 5
A Survey of Confidence Estimation and Calibration in Large Language ModelsPreprint
Geng, J., Cai, F., Wang, Y., Koeppl, H., Nakov, P., & Gurevych, I. (2023) · arXiv (Cornell University) · doi:10.48550/arxiv.2311.08298 · Cited by 4
What Large Language Models Know and What People Think They KnowPreprint
Belem, C., Lemus, H. T., & Steyvers, M. (2024) · OSF Preprints (OSF Preprints) · doi:10.17605/osf.io/y7pr6 · Cited by 1
Self-citationOpenAI, Achiam, J., Adler, S., Agarwal, S., Ahmad, L., Akkaya, I., Aleman, F. L., Almeida, D., Altenschmidt, J., Altman, S., et al. (2023) · arXiv · doi:10.48550/arxiv.2303.08774
Self-citation
Research works printed but not linked: 9
- Karelitz, T. M., Dhami, M. K., Budescu, D. V. & Wallsten, T. S. Toward a universal translator of verbal probabilities. In Proc. 15th International Florida Artificial Intelligence Research Society Conference (eds Haller, M. S. & Simmons, G.) 498–502 (AAAI Press, 2002).
- O’Brien, B. J. Words or numbers? The evaluation of probability expressions in general practice. J. R. Coll. Gen. Pract. 39, 98–100 (1989).
- Whalen, J. et al. ChatGPT: challenges, opportunities, and implications for teacher education. Contemp. Iss. Technol. Teach. Educ. 23, 1–23 (2023).
- Azaria, A. & Mitchell, T. in Findings of the Association for Computational Linguistics: EMNLP 2023 (eds Bouamor, H. et al.) 967–976 (ACL, 2023).
- Xiao, Y. et al. in Findings of the Association for Computational Linguistics: EMNLP 2022 (eds Goldberg, Y. et al.) 7273–7284 (ACL, 2022).
- Tanneru, S. H., Agarwal, C. & Lakkaraju, H. Quantifying uncertainty in natural language explanations of large language models. In International Conference on Artificial Intelligence and Statistics 1072–1080 (PMLR, 2024).
- Lin, S., Hilton, J. & Evans, O. Teaching models to express their uncertainty in words. Trans. Mach. Learn. Res. (2022).
- Guo, C., Pleiss, G., Sun, Y. & Weinberger, K. Q. On calibration of modern neural networks. In Proc. 34th International Conference on Machine Learning (eds Precup, D. & Teh, Y. W.) vol. 70 of Proceedings of Machine Learning Research 1321–1330 (PMLR, 2017).
- Kumar, A., Liang, P. S. & Ma, T. Verified uncertainty calibration. Adv. Neural Inf. Process. Syst. 32, (2019).
Cited by
A cognitive approach to human–AI complementarity in dynamic decision-making
González, C., & Heidari, H. (2025) · Nature Reviews Psychology · doi:10.1038/s44159-025-00499-x · Cited by 92
LLM-Based Multi-Agent Decision-Making: Challenges and Future Directions
Sun, C., Huang, S., & Pompili, D. (2025) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2025.3562371 · Cited by 102
Generative Deep Learning for de Novo Drug Design─A Chemical Space Odyssey
Özçelik, R., Brinkmann, H., Criscuolo, E., & Grisoni, F. (2025) · Journal of Chemical Information and Modeling · doi:10.1021/acs.jcim.5c00641 · Cited by 71
The homogenizing effect of large language models on human expression and thought
Sourati, Z., Ziabari, A. S., & Dehghani, M. (2026) · Trends in Cognitive Sciences · doi:10.1016/j.tics.2026.01.003 · Cited by 127
Tiller, N. B., Marcon, A. R., Zenone, M. A., Kidd, K. E., Jeukendrup, A. E., Master, Z., & Caulfield, T. (2026) · BMJ Open · doi:10.1136/bmjopen-2025-112695 · Cited by 58
Metadata
- Authors
- Steyvers, M., Tejeda, H., Kumar, A., Belem, C., Karny, S., Hu, X., Mayer, L. W., & Smyth, P.
- Affiliations
- University of California, Irvine (US)
- Corresponding author
- Steyvers, Mark
- ORCID
- Venue
- Nature Machine Intelligence
- Published
- 2025-01-21
- Volume
- 7
- Issue
- 2
- Pages
- 221-231
- ISSN
- 2522-5839 (ISSN-L)
- OpenAlex
- W4406679533
- Cited by
- 209
- References
- 36
- Citations by year
- 2026: 147 · 2025: 60
- FWCI
- 83.03
- Type
- journal article
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- cc-by
- creativecommons.org/licenses/by/4.0 (tdm)
- creativecommons.org/licenses/by/4.0 (vor)
- Locations
- Nature Machine Intelligence (publishedVersion, cc-by)
- arXiv (Cornell University) (submittedVersion, cc-by-sa)
- arXiv (Cornell University)
- Funders
- National Science Foundation (1900644)
- Hasso Plattner Institute (HPI) Research Center in Machine Learning and Data Science at the University of California, Irvine
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Explainable Artificial Intelligence (XAI)
- Physical Sciences › Computer Science › Artificial Intelligence › Ethics and Social Impacts of AI
- Social Sciences › Human Factors and Ergonomics › Human-Automation Interaction and Safety
- Keywords
- large language models, uncertainty communication, user confidence, trust in AI, AI-assisted decision-making, confidence calibration, human-AI interaction
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.