The DOI matches its Crossref record; 16 of the 16 references it cites were found in a registry or an index.
Human bias in AI models? Anchoring effects and mitigation strategies in large language models
Abstract
This study builds on the seminal work of Tversky and Kahneman (1974), exploring the presence and extent of anchoring bias in forecasts generated by four Large Language Models (LLMs): GPT-4, Claude 2, Gemini Pro and GPT-3.5. In contrast to recent findings of advanced reasoning capabilities in LLMs, our randomised controlled trials reveal the presence of anchoring bias across all models: forecasts are significantly influenced by prior mention of high or low values. We examine two mitigation prompting strategies, ‘Chain of Thought’ and ‘ignore previous’, finding limited and varying degrees of effectiveness. Our results extend the anchoring bias research in finance beyond human decision-making to encompass LLMs, highlighting the importance of deliberate and informed prompting in AI forecasting in both ad hoc LLM use and in crafting few-shot examples.
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.
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?
In the citing paper
“Reasoning through these requires causal inference, temporal sequencing, and some skepticism-capabilities not captured by endpoint accuracy 45,46 .”
From Structured reasoning failures compromise LLM interpretation of clinical oncology notes (2026) · cites this paper as [45]
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
Tversky, A., & Kahneman, D. (1974) · Science · doi:10.1126/science.185.4157.1124 · Cited by 28,810
On the Dangers of Stochastic Parrots: Can Language Models Be Too Big? 🦜
Bender, E. M., Gebru, T., McMillan-Major, A., & Shmitchell, S. (2021) · Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency · doi:10.1145/3442188.3445922 · Cited by 6,988
Chain-Of-Thought Prompting Elicits Reasoning in Large Language Models
Wei, J., Wang, X., Schuurmans, D., Bosma, M., Ichter, B., Xia, F., Chi, E., Le, Q. V., & Zhou, D. (2022) · Advances in Neural Information Processing Systems 35 · doi:10.52202/068431-1800 · Cited by 4,505
Large language models are zero-shot reasoners
Kojima, T., Gu, S. (., Reid, M., Matsuo, Y., & Iwasawa, Y. (2022) · Advances in Neural Information Processing Systems 35 · doi:10.52202/068431-1613 · Cited by 2,013
Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena
Zheng, L., Chiang, W.-L., Sheng, Y., Zhuang, S., Wu, Z., Zhuang, Y., Lin, Z., Li, Z., Li, D., Xing, E. P., et al. (2023) · arXiv · doi:10.48550/arxiv.2306.05685 · Cited by 1,888
Dell'Acqua, F., McFowland III, E., Mollick, E. R., Lifshitz-Assaf, H., Kellogg, K., Rajendran, S., Krayer, L., Candelon, F., & Lakhani, K. R. (2023) · SSRN Electronic Journal · doi:10.2139/ssrn.4573321 · Cited by 794
Using cognitive psychology to understand GPT-3
Binz, M., & Schulz, E. (2023) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.2218523120 · Cited by 559
Englich, B., Mussweiler, T., & Strack, F. (2006) · Personality and Social Psychology Bulletin · doi:10.1177/0146167205282152 · Cited by 554
Can ChatGPT forecast stock price movements? Return predictability and large language models
Lopez-Lira, A., & Tang, Y. (2026) · Journal of Financial Economics · doi:10.1016/j.jfineco.2026.104335 · Cited by 425
Evaluating Large Language Models in Theory of Mind Tasks
Kosinski, M. (2023) · arXiv · doi:10.48550/arxiv.2302.02083 · Cited by 339
Hagendorff, T., Fabi, S., & Kosinski, M. (2023) · Nature Computational Science · doi:10.1038/s43588-023-00527-x · Cited by 243
The anchoring bias reflects rational use of cognitive resources
Lieder, F., Griffiths, T. L., M. Huys, Q. J., & Goodman, N. D. (2018) · Psychonomic Bulletin & Review · doi:10.3758/s13423-017-1286-8 · Cited by 230
The emergence of economic rationality of GPT
Chen, Y., Liu, T. X., Shan, Y., & Zhong, S. (2023) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.2316205120 · Cited by 164
A Manager and an AI Walk into a Bar: Does ChatGPT Make Biased Decisions Like We Do?
Chen, Y., Kirshner, S. N., Ovchinnikov, A., Andiappan, M., & Jenkin, T. (2025) · Manufacturing & Service Operations Management · doi:10.1287/msom.2023.0279 · Cited by 143
Ignore Previous Prompt: Attack Techniques For Language Models
Perez, F., & Ribeiro, I. (2022) · arXiv · doi:10.48550/arxiv.2211.09527 · Cited by 110
The Reversal Curse: LLMs trained on "A is B" fail to learn "B is A"
Berglund, L., Tong, M., Kaufmann, M., Balesni, M., Stickland, A. C., Korbak, T., & Evans, O. (2023) · arXiv · doi:10.48550/arxiv.2309.12288 · Cited by 36
Research works printed but not linked: 24
- Akter, 2021, Algorithmic bias in data-driven innovation in the age of AI, Int. J. Inf. Manag., 60, 10.1016/j.ijinfomgt.2021.102387 · DOI · Paper page
- Ashrafimoghari, V., Gürkan, N., Suchow, J.W., 2024. Evaluating Large Language Models on the GMAT: Implications for the Future of Business Education. https://doi.org/10.48550/arXiv.2401.02985. · arXiv · doi:10.2139/ssrn.4681307
- Belot, 2023, Australian academics apologise for false AI-generated allegations against big four consultancy firms, Guardian
- Cen, 2013, The role of anchoring bias in the equity market: evidence from analysts’ earnings forecasts and stock returns, J. Financ. Quant. Anal., 48, 47, 10.1017/S0022109012000609 · DOI · Paper page
- Chu, 2023, A survey of chain of thought reasoning: advances, Front. Future
- Coda-Forno, 2023, Inducing Anxiety Large Lang. Models Increases Explor. bias
- Dowling, 2023, ChatGPT for (Finance) research: the bananarama conjecture, Financ. Res. Lett., 53, 10.1016/j.frl.2023.103662 · DOI · Paper page
- Goodell, 2021, Artificial intelligence and machine learning in finance: identifying foundations, themes, and research clusters from bibliometric analysis, J. Behav. Exp. Financ., 32, 10.1016/j.jbef.2021.100577 · DOI · Paper page
- Gray, 2023, Measurement and mitigation of bias in artificial intelligence: a narrative literature review for regulatory science, Clin. Pharmacol. Ther.
- Gurdgiev, 2020, Herding and anchoring in cryptocurrency markets: investor reaction to fear and uncertainty, J. Behav. Exp. Financ., 25, 10.1016/j.jbef.2020.100271 · DOI · Paper page
- Gurnee, 2023, Lang. Models Represent Space Time
- Harvey, 2007, Use of heuristics: Insights from forecasting research, Think. Reason., 13, 5, 10.1080/13546780600872502 · DOI
- Hasan, 2022, Can Artificial Intelligence (AI) manage behavioural biases among financial planners?, J. Glob. Inf. Manag. JGIM, 31, 1
- Hendy, 2021, Unsticking credit card repayments from the minimum: advice, anchors and financial incentives, J. Behav. Exp. Financ., 30, 10.1016/j.jbef.2021.100505 · DOI
- Jacowitz, 1995, Measures of anchoring in estimation tasks, Pers. Soc. Psychol. Bull., 21, 1161, 10.1177/01461672952111004 · DOI · Paper page
- Königstorfer, 2020, Applications of Artificial Intelligence in commercial banks – a research agenda for behavioral finance, J. Behav. Exp. Financ., 27, 10.1016/j.jbef.2020.100352 · DOI · Paper page
- Koralus, P., Wang-Maścianica, V., 2023. Humans in Humans Out: On GPT Converging Toward Common Sense in both Success and Failure [WWW Document]. arXiv.org. URL 〈https://arxiv.org/abs/2303.17276v1〉 (accessed 2.7.24). · arXiv
- Li, 2023, Can ChatGPT reduce human financial analysts’ optimistic biases?, Econ. Polit. Stud., 0, 1
- Løhre, 2016, Numerical anchors and their strong effects on software development effort estimates, J. Syst. Softw., 116, 49, 10.1016/j.jss.2015.03.015 · DOI
- Meub, 2016, Can anchoring explain biased forecasts? Experimental evidence, J. Behav. Exp. Financ., 12, 1, 10.1016/j.jbef.2016.08.001 · DOI
- Milmo, 2023, Two US lawyers fined for submitting fake court citations from ChatGPT, Guardian
- Opedal, A., Stolfo, A., Shirakami, H., Jiao, Y., Cotterell, R., Schölkopf, B., Saparov, A., Sachan, M., 2024. Do Language Models Exhibit the Same Cognitive Biases in Problem Solving as Human Learners? https://doi.org/10.48550/arXiv.2401.18070. · DOI
- Shanmuganathan, 2020, Behavioural finance in an era of artificial intelligence: longitudinal case study of robo-advisors in investment decisions, J. Behav. Exp. Financ., 27, 10.1016/j.jbef.2020.100297 · DOI · Paper page
- Talboy, A.N., Fuller, E., 2023. Challenging the appearance of machine intelligence: Cognitive bias in LLMs. https://doi.org/10.48550/arXiv.2304.01358. · DOI
Cited by
Anchoring Bias in Large Language Models: An Experimental Study
Lou, J., & Sun, Y. (2025) · doi:10.32388/o2n5j6 · Cited by 5
Structured reasoning failures compromise LLM interpretation of clinical oncology notes
Kenaston, M. W., Ayub, U., Parmar, M., Kumar, P., Anjum, M. U., Naqvi, S. A. A., Chaudhuri, A. A., Zakharia, Y., Heath, E. I., Bekaii-Saab, T. S., et al. (2026) · npj Digital Medicine · doi:10.1038/s41746-026-02951-5 · Cited by 2
Metadata
- Authors
- Nguyen, J. K.
- Venue
- Journal of Behavioral and Experimental Finance
- Publisher
- Elsevier BV
- Published
- 2024-09
- Volume
- 43
- Pages
- 100971
- ISSN
- 2214-6350
- Cited by
- 33
- References
- 16
- Language
- en
- Licence
- www.elsevier.com/tdm/userlicense/1.0/ (tdm)
- www.elsevier.com/legal/tdmrep-license (tdm)
- creativecommons.org/licenses/by/4.0/ (vor)
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.