On the Dangers of Stochastic Parrots: Can Language Models Be Too Big? 🦜
Open accessThe DOI matches its Crossref record; 156 of the 159 references it cites were found in a registry or an index, and 3 are not checked yet.
Abstract
The past 3 years of work in NLP have been characterized by the development and deployment of ever larger language models, especially for English. BERT, its variants, GPT-2/3, and others, most recently Switch-C, have pushed the boundaries of the possible both through architectural innovations and through sheer size. Using these pretrained models and the methodology of fine-tuning them for specific tasks, researchers have extended the state of the art on a wide array of tasks as measured by leaderboards on specific benchmarks for English. In this paper, we take a step back and ask: How big is too big? What are the possible risks associated with this technology and what paths are available for mitigating those risks? We provide recommendations including weighing the environmental and financial costs first, investing resources into curating and carefully documenting datasets rather than ingesting everything on the web, carrying out pre-development exercises evaluating how the planned approach fits into research and development goals and supports stakeholder values, and encouraging research directions beyond ever larger language models.
Checked 2026-11-22 · 3 references not checked yet
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The text we found for this paper looks like another document, so the self-check wasn’t run.
Citations
Cites
Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, L., & Polosukhin, I. (2017) · arXiv · doi:10.48550/arxiv.1706.03762 · Cited by 120,404
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. (2019) · Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers) · doi:10.18653/v1/n19-1423 · Cited by 54,325 · cited 5×
“[39] in particular noted that training on a large dataset and fine-tuning for specific tasks leads to strictly increasing results on the GLUE tasks [138] for English as the hyperparameters of the model were increased.”
BackgroundMethodResultInfluentialGlove: Global Vectors for Word Representation
Pennington, J., Socher, R., & Manning, C. (2014) · Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.3115/v1/d14-1162 · Cited by 34,071
Language Models are Few-Shot Learners
Brown, T. B., Mann, B., Ryder, N., Subbiah, M., Kaplan, J., Dhariwal, P., Neelakantan, A., Shyam, P., Sastry, G., Askell, A., et al. (2020) · arXiv · doi:10.48550/arxiv.2005.14165 · Cited by 23,372 · cited 4×
“documents linked to from Reddit [25], plus Wikipedia and a collection”
BackgroundMethodInfluentialRoBERTa: A Robustly Optimized BERT Pretraining Approach
Liu, Y., Ott, M., Goyal, N., Du, J., Joshi, M., Chen, D., Levy, O., Lewis, M., Zettlemoyer, L., & Stoyanov, V. (2019) · arXiv · doi:10.48550/arxiv.1907.11692 · Cited by 16,515
Distilling the Knowledge in a Neural Network
Hinton, G., Vinyals, O., & Dean, J. (2015) · arXiv · doi:10.48550/arxiv.1503.02531 · Cited by 14,204 · cited 1×
“[26, 58], quantization [118, 153], factorized embedding parameterization and cross-layer parameter sharing [70], and progressive module replacing [146].”
Deep Contextualized Word Representations
Peters, M., Neumann, M., Iyyer, M., Gardner, M., Clark, C., Lee, K., & Zettlemoyer, L. (2018) · Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long Papers) · doi:10.18653/v1/n18-1202 · Cited by 11,983
Proceedings of the 24th international conference on Machine learning
(2007) · doi:10.1145/1273496 · Cited by 11,630
Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer
Raffel, C., Shazeer, N., Roberts, A., Lee, K., Narang, S., Matena, M., Zhou, Y., Li, W., & Liu, P. J. (2019) · arXiv · doi:10.48550/arxiv.1910.10683 · Cited by 10,929
The Mathematical Theory of Communication
Shannon, C. E., Weaver, W., & Wiener, N. (1950) · Physics Today · doi:10.1063/1.3067010 · Cited by 9,801
Research works printed but not linked: 48
- [4] American Psychological Association. 2019. Discrimination: What it is, and how to cope. https://www.apa.org/topics/discrimination (2019). Google Scholar
- 10.18653/v1 · DOI
- 10.18653/v1 · DOI
- [15] Ruha Benjamin. 2019. Race After Technology: Abolitionist Tools for the New Jim Code. Polity Press, Cambridge, UK. Crossref Google Scholar
- [16] Elettra Bietti and Roxana Vatanparast. 2020. Data Waste. Harvard International Law Journal 61 (2020). Google Scholar
- 10.18653/v1 · DOI
- 10.18653/v1 · DOI
- 10.1016/0010-0277(86) · doi:10.1016/0010-0277(86
- [38] Terrance de Vries, Ishan Misra, Changhan Wang, and Laurens van der Maaten. 2019. Does object recognition work for everyone?. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops. 52--59. Google Scholar
- 10.18653/v1 · DOI
- [42] Ethan Fast, Tina Vachovsky, and Michael Bernstein. 2016. Shirtless and Dangerous: Quantifying Linguistic Signals of Gender Bias in an Online Fiction Writing Community. In Proceedings of the International AAAI Conference on Web and Social Media, Vol. 10. Google Scholar
- 10.18653/v1 · DOI
- [45] Darja Fišer, Ruihong Huang, Vinodkumar Prabhakaran, Rob Voigt, Zeerak Waseem, and Jacqueline Wernimont (Eds.). 2018. Proceedings of the 2nd Workshop on Abusive Language Online (ALW2). Association for Computational Linguistics, Brussels, Belgium. https://www.aclweb.org/anthology/W18-5100 Google Scholar
- [50] Batya Friedman, Peter H. Kahn, Jr., and Alan Borning. 2006. Value sensitive design and information systems. In Human-Computer Interaction in Management Information Systems: Foundations, P Zhang and D Galletta (Eds.). M. E. Sharpe, Armonk NY, 348--372. Google Scholar
- 10.18653/v1 · DOI
- 10.18653/v1 · DOI
- 10.18653/v1 · DOI
- [67] Brendan Kennedy, Drew Kogon, Kris Coombs, Joseph Hoover, Christina Park, Gwenyth Portillo-Wightman, Aida Mostafazadeh Davani, Mohammad Atari, and Morteza Dehghani. 2018. A typology and coding manual for the study of hate-based rhetoric. PsyArXiv. July 18 (2018). Google Scholar
- [77] Marianna Martindale and Marine Carpuat. 2018. Fluency Over Adequacy: A Pilot Study in Measuring User Trust in Imperfect MT. In Proceedings of the 13th Conference of the Association for Machine Translation in the Americas (Volume 1: Research Track). Association for Machine Translation in the Americas, Boston, MA, 13--25. https://www.aclweb.org/anthology/W18-1803 Google Scholar
- [81] Douglas M McLeod. 2007. News coverage and social protest: How the media's protect paradigm exacerbates social conflict. Journal of Dispute Resolution (2007), 185. Google Scholar
- [87] Robert C. Moore and William Lewis. 2010. Intelligent Selection of Language Model Training Data. In Proceedings of the ACL 2010 Conference Short Papers. Association for Computational Linguistics, Uppsala, Sweden, 220--224. https://www.aclweb.org/anthology/P10-2041 Google Scholar
- 10.18653/v1 · DOI
- [92] Maggie Nelson. 2015. The Argonauts. Graywolf Press, Minneapolis. Google Scholar
- 10.18653/v1 · DOI
- [96] David Ortiz, Daniel Myers, Eugene Walls, and Maria-Elena Diaz. 2005. Where do we stand with newspaper data? Mobilization: An International Quarterly 10, 3 (2005), 397--419. Google Scholar
- 10.3115/v1 · doi:10.3115/v1
- 10.18653/v1 · DOI
- 10.18653/v1 · DOI
- [106] Alec Radford, Jeffrey Wu, Rewon Child, David Luan, Dario Amodei, and Ilya Sutskever. 2019. Language models are unsupervised multitask learners. OpenAI Blog 1, 8 (2019), 9. Google Scholar
- 10.18653/v1 · DOI
- [109] Sarah T. Roberts, Joel Tetreault, Vinodkumar Prabhakaran, and Zeerak Waseem (Eds.). 2019. Proceedings of the Third Workshop on Abusive Language Online. Association for Computational Linguistics, Florence, Italy. https://www.aclweb.org/anthology/W19-3500 Google Scholar
- [112] Corby Rosset. 2020. Turing-NLG: A 17-billion-parameter language model by Microsoft. Microsoft Blog (2020). Google Scholar
- 10.18653/v1 · DOI
- 10.18653/v1 · DOI
- [121] Joonbo Shin, Yoonhyung Lee, and Kyomin Jung. 2019. Effective Sentence Scoring Method Using BERT for Speech Recognition. In Asian Conference on Machine Learning. 1081--1093. Google Scholar
- [124] Karen Spärck Jones. 2004. Language modelling's generative model: Is it rational? Technical Report. Computer Laboratory, University of Cambridge. Google Scholar
- [127] Katrina Srigley and Lorraine Sutherland. 2019. Decolonizing, Indigenizing, and Learning Biskaaybiiyang in the Field: Our Oral History Journey1. The Oral History Review (2019). Google Scholar
- 10.18653/v1 · DOI
- 10.18653/v1 · DOI
- 10.18653/v1 · DOI
- 10.18653/v1 · DOI
- 10.18653/v1 · DOI
- 10.18653/v1 · DOI
- 10.18653/v1 · DOI
- [150] Zhilin Yang, Zihang Dai, Yiming Yang, Jaime Carbonell, Russ R Salakhutdinov, and Quoc V Le. 2019. Xlnet: Generalized autoregressive pretraining for language understanding. In Advances in neural information processing systems. 5753--5763. Google Scholar
- 10.18653/v1 · DOI
- 10.18653/v1 · DOI
- 10.18653/v1 · DOI
Cited by
Items 1,301–1,400 of 6,931
A Unified Framework for Assessing Energy Efficiency of Machine Learning
Fischer, R., Jakobs, M., Mücke, S., & Morik, K. (2023) · Communications in Computer and Information Science · doi:10.1007/978-3-031-23618-1_3 · Cited by 16
Investigating Layer Importance in Large Language Models
Zhang, Y., Dong, Y., & Kawaguchi, K. (2024) · Proceedings of the 7th BlackboxNLP Workshop: Analyzing and Interpreting Neural Networks for NLP · doi:10.18653/v1/2024.blackboxnlp-1.29 · Cited by 16
Meinlschmidt, G., Koc, S., Boerner, E., Tegethoff, M., Simacek, T., Schirmer, L., & Schneider, M. (2025) · BMC Medical Education · doi:10.1186/s12909-025-07307-3 · Cited by 15
“First, the authenticity of interactions can be limited if the AI’s generated text fails to capture the nuance of human responses or emotional expressions [12].”
In conversation with Artificial Intelligence: aligning language models with human values
Kasirzadeh, A., & Gabriel, I. (2022) · arXiv · doi:10.48550/arxiv.2209.00731 · Cited by 128
Usage-based constructionist approaches and large language models
Goldberg, A. E. (2024) · Constructions and Frames · doi:10.1075/cf.23017.gol · Cited by 57
“Many have rightly observed that there is a dark side to successful language models that ought not be ignored (e.g., Bender et al. 2021; Klein 2003; van Dis et al. 2023).”
BackgroundWarner, M., Strohmayer, A., Higgs, M., & Coventry, L. (2025) · International Journal of Human-Computer Studies · doi:10.1016/j.ijhcs.2025.103468 · Cited by 31
Allan, K., Azcona, J., Sripada, S. G., Leontidis, G., Sutherland, C. A. M., Phillips, L. H., & Martin, D. S. (2025) · Royal Society Open Science · doi:10.1098/rsos.241472 · Cited by 28
“After training, LLMs need to be prompted, i.e. provided with an input stream of ‘tokens’ to produce an output [69].”
Classifying legal interpretations using large language models
Dugac, G., & Altwicker, T. (2025) · Artificial Intelligence and Law · doi:10.1007/s10506-025-09447-9 · Cited by 30
Cut the crap: a critical response to “ChatGPT is bullshit”
Gunkel, D. J., & Coghlan, S. (2025) · Ethics and Information Technology · doi:10.1007/s10676-025-09828-3 · Cited by 54
Autoethnographic Insights from Neurodivergent GAI “Power Users”
Glazko, K. S., Cha, J., Lewis, A., Kosa, B., Wimer, B. L., Zheng, A., Zheng, Y., & Mankoff, J. C. (2025) · CHI Conference on Human Factors in Computing Systems · doi:10.1145/3706598.3713670 · Cited by 32
Fan, X., Xiao, Q., Zhou, X., Pei, J., Sap, M., Lu, Z., & Shen, H. (2025) · CHI Conference on Human Factors in Computing Systems · doi:10.1145/3706598.3713477 · Cited by 66
Kong, H., Lowy, R., Choi, Y., & Kim, J. G. (2025) · CHI Conference on Human Factors in Computing Systems · doi:10.1145/3706598.3713344 · Cited by 23
The bewitching AI: The Illusion of Communication with Large Language Models
Bottazzi Grifoni, E., & Ferrario, R. (2025) · Philosophy & Technology · doi:10.1007/s13347-025-00893-6 · Cited by 15
“Characterizing them as “stochastic parrots” (Bender et al., 2021) does not do justice to their capabilities because, while though LLMs are skilled at recognizing recurring speech patterns (that are somewhat stereotypical), they also introduce a form of variability (Atil et al., 2025) in their…”
Recommender Systems Meet Large Language Model Agents: A Survey
Zhu, X., Wang, Y., Gao, H., Xu, W., Wang, C., Liu, Z., Wang, K., Jin, M., Pang, L. X., Weng, Q., et al. (2025) · Foundations and Trends® in Privacy and Security · doi:10.1561/3300000050 · Cited by 47
Exploring the ethical landscape of AI in academic writing
Ateriya, N., Sonwani, N. S., Thakur, K. S., Kumar, A., & Verma, S. K. (2025) · Egyptian Journal of Forensic Sciences · doi:10.1186/s41935-025-00453-1 · Cited by 47
Can ChatGPT enhance business student creativity? Evidence from a randomised controlled trial
Fisher, R., Gogan, T. D., Williams, J. S., Laferriere, R., Campbell, G., Gunasekara, A. N., & Nguyen, J. (2025) · Studies in Higher Education · doi:10.1080/03075079.2025.2515512 · Cited by 23
“…lack agency (i.e. they do not display intention, cannot understand their actions, do not apply judgement, and lack evaluative ability), the output of LLMs does not meet the conditions to be described creative and are better understood as purely algorithmic ‘stochastic parrots’ (Bender et al. 2021).”
The disruption of grief in the technological niche: The case of deathbots
Fabry, R. E. (2025) · Phenomenology and the Cognitive Sciences · doi:10.1007/s11097-025-10083-6 · Cited by 27
“Finally, LLM-based chatbots, including deathbots, are not able to understand linguistic inputs and to relate to the cognitive-affective niche through linguistic outputs in any meaningful way (Bender et al., 2021).”
Considering the Social and Economic Sustainability of AI
Waelen, R. A., & van Wynsberghe, A. (2025) · Science and Engineering Ethics · doi:10.1007/s11948-025-00544-1 · Cited by 22
LLMs outperform outsourced human coders on complex textual analysis
Bermejo, V. J., Gago, A., Gálvez, R. H., & Harari, N. (2025) · Scientific Reports · doi:10.1038/s41598-025-23798-y · Cited by 40
“These risks are further exacerbated by the opaque nature of model reasoning, which limits interpretability and raises accountability concerns in sensitive domains such as sociopolitical classification 90 .”
Bojić, L., Ilić, V., Prodanović, V., & Vuković, V. (2025) · Chinese Political Science Review · doi:10.1007/s41111-025-00326-x · Cited by 25
RuleBERT: Teaching Soft Rules to Pre-Trained Language Models
Saeed, M., Ahmadi, N., Nakov, P., & Papotti, P. (2021) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2021.emnlp-main.110 · Cited by 44
Proportionality Principle for the Ethics of Artificial Intelligence
Karliuk, M. (2022) · AI and Ethics · doi:10.1007/s43681-022-00220-1 · Cited by 26
Energy and Space Efficient Parallel Adder Using Molecular Memristors
Yi, S., Rath, S. P., ., D., Venkatesan, T., Bhat, N., Goswami, S., Williams, R. S., & Goswami, S. (2022) · Advanced Materials · doi:10.1002/adma.202206128 · Cited by 15
Contrasting Technologists’ and Activists’ Positions on Signing Avatars
Angelini, R. (2023) · CHI Conference Extended Abstracts on Human Factors in Computing Systems (CHI EA) · doi:10.1145/3544549.3583946 · Cited by 20
Kikerpill, K., & Siibak, A. (2023) · Computers in the Schools · doi:10.1080/07380569.2023.2244941 · Cited by 22
“In fact, critical voices have previously referred to the output of language models like GPT-3 as “sto-chastic parrots” (Bender et al., 2021), indicating that the responses of such systems “echo what they hear, remixed by randomness” (Hutson, 2021, 23).”
BackgroundHacking by the prompt: Innovative ways to utilize ChatGPT for evaluators
Ferretti, S. (2023) · New Directions for Evaluation · doi:10.1002/ev.20557 · Cited by 20
AI research ethics is in its infancy: the EU’s AI Act can make it a grown-up
Rességuier, A., & Ufert, F. (2023) · Research Ethics · doi:10.1177/17470161231220946 · Cited by 29
“…data (Weitzenboeck et al., 2022); challenges related to seeking consent (Andreotta et al., 2021; Zimmer, 2018); data quality and curation challenges (Bender et al., 2021; Birhane and Prabhu, 2021); and issues related to the labelling process (Scheuerman et al., 2019), outsourcing of this process…”
BackgroundAlgorithmic Ventriloquism: The Contested State of Voice in AI Speech Generators
Ramati, I. (2024) · Social Media + Society · doi:10.1177/20563051231224401 · Cited by 22
“…suggests that they are not simply “stochastic parrots” which imitate human activity, working as “systems for haphazardly stitching together sequences of linguistic forms . . . according to probabilistic information . . . but without any reference to meaning” (Bender et al., 2021, p. 617).”
BackgroundReal sparks of artificial intelligence and the importance of inner interpretability
Grzankowski, A. (2024) · Inquiry · doi:10.1080/0020174x.2023.2296468 · Cited by 29
“So unless we think this time is very different, we shouldn’t conclude that we are dealing with anything but another mimic or, as Bender et al (2021) put it, with a ‘stochastic parrot’.”
BackgroundDeciphering Deception: How Different Rhetoric of AI Language Impacts Users’ Sense of Truth in LLMs
Yoo, D., Kang, H., & Oh, C. (2024) · International Journal of Human-Computer Interaction · doi:10.1080/10447318.2024.2316370 · Cited by 28
“Instead of constructing answers based on verified facts, they operate probabilistically, selecting words based on their likelihood to follow previous words, and as such (Bender et al., 2021) inherently don’t distinguish between truth and falsehood (Danry et al., 2022).”
Loos, E. F., & Radicke, J. (2024) · AI and Ethics · doi:10.1007/s43681-024-00470-1 · Cited by 19
Omizo, R. M. (2024) · Journal of Business and Technical Communication · doi:10.1177/10506519241239927 · Cited by 17
“…conclusions inductively from LLM output, given their current inability to make knowledgeable inferences about novel styles, ideas, or organization (Bender et al., 2021), the fi ndings of this study suggest that for many of the regularized genres employed in BTC, such as technical or product…”
BackgroundOn Causality in Distributed Continuum Systems
Pujol, V. C., Sedlak, B., Donta, P. K., & Dustdar, S. (2024) · IEEE Internet Computing · doi:10.1109/mic.2023.3344248 · Cited by 22
“Although all new capacities are discovered and enhanced through extensive training, many inconvenient aspects of this technology, such as a lack of trustworthiness 6 or being resource devouring, 7 have only emerged since ML/AI have become essential in our society.”
BackgroundPitfalls (and advantages) of sophisticated large language models
Strasser, A. (2024) · Elsevier eBooks · doi:10.1016/b978-0-443-18851-0.00007-x · Cited by 20
VAL: Interactive Task Learning with GPT Dialog Parsing
Lawley, L., & MacLellan, C. J. (2024) · CHI Conference on Human Factors in Computing Systems · doi:10.1145/3613904.3641915 · Cited by 33
“Additionally, the models we used are not open-source, and the centralization of foundational technology behind the veil of a corporate API offers almost no transparency, both into how the technology works and into how the data sent by users is being used or stored [7].”
MethodNatural language processing: An overview of models, transformers and applied practices
Canchila, S., Meneses-Eraso, C., Casanoves-Boix, J., Cortés-Pellicer, P., & Sirvent, F. C. (2024) · Computer Science and Information Systems · doi:10.2298/csis230217031c · Cited by 31
Katirai, A., García, N., Ide, K., Nakashima, Y., & Kishimoto, A. (2024) · AI and Ethics · doi:10.1007/s43681-024-00517-3 · Cited by 26
“…et al. [51] and Ko et al. [52] and with acknowledgement of recent work on image generation models [10, 53] and reference to recent work on LLMs [12, 54], we present seven issue clusters: data issues, intellectual property, bias, privacy, and the impacts on the informational, cultural, and natural…”
Abernethy, N. J. (2024) · AI and Ethics · doi:10.1007/s43681-024-00575-7 · Cited by 20
“13 This is because LLMs are biased towards these views and paradigms, given that these things make up the majority of their training data [33].”
InfluentialBelcher, D. (2024) · Journal of English for Research Publication Purposes · doi:10.1075/jerpp.00025.bel · Cited by 18
LLM-mediated domain-specific voice agents: the case of TextileBot
Zhong, S., Gatti, E., Hardwick, J., Ribul, M., Cho, Y., & Obrist, M. (2025) · Behaviour and Information Technology · doi:10.1080/0144929x.2025.2456667 · Cited by 22
Artificial intelligence and the death of the academic author
Masters, K. (2025) · Medical Teacher · doi:10.1080/0142159x.2025.2523463 · Cited by 17
Towards Computational Support with Language Models for TTRPG Game Masters
Kelly, J., Mateas, M., & Wardrip–Fruin, N. (2023) · International Conference on the Foundations of Digital Games (FDG) · doi:10.1145/3582437.3587202 · Cited by 21
Hamid, O. H. (2023) · International Conference on Information Technology Trends (ITT) · doi:10.1109/itt59889.2023.10184233 · Cited by 16
“fact that GPTs primarily rely on statistical patterns and lack an inherent grasp of the underlying meaning within the language they process [21].”
BackgroundEvaluating Embedding APIs for Information Retrieval
Kamalloo, E., Zhang, X., Ogundepo, O., Thakur, N., Alfonso-Hermelo, D., Rezagholizadeh, M., & Lin, J. (2023) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2023.acl-industry.50 · Cited by 37
Global and Local Hierarchy-aware Contrastive Framework for Implicit Discourse Relation Recognition
Jiang, Y., Zhang, L., & Wang, W. (2023) · Findings of the Association for Computational Linguistics: ACL (Findings of ACL) · doi:10.18653/v1/2023.findings-acl.510 · Cited by 33
Exploring the Impacts of ChatGPT on Future Scientific WorkPreprint
Irons, J. L., Mason, C., Cooper, P. S., Sidra, S., Reeson, A. F., & Paris, C. L. (2023) · SocArXiv (OSF Preprints) · doi:10.31235/osf.io/j2u9x · Cited by 18
Identification and Description of Emotions by Current Large Language ModelsPreprint
Patel, S., & Fan, J. (2023) · Research Square · doi:10.1101/2023.07.17.549421 · Cited by 27
“Experts acknowledge that there are risks with scaling large language models even further [78] and that there lies responsibility within the technical community to develop products with a process that mitigates harmful consequence and promote human well-being by adopting responsible computing…”
BackgroundPersonal experience with AI-generated peer reviews: a case study
Lo Vecchio, N. (2025) · Research Integrity and Peer Review · doi:10.1186/s41073-025-00161-3 · Cited by 15
Large language models in peer review: challenges and opportunities
Sun, Z. (2025) · Scientometrics · doi:10.1007/s11192-025-05440-w · Cited by 15
“This lack of transparency can result in misunderstandings and misuse of LLM-generated outputs, particularly within academic publishing and peer review contexts (Bender et al., 2021; ICLR, 2025).”
A comparative study on the impact of model compression techniques on fairness in language models
Ramesh, K., Chavan, A., Pandit, S., & Sitaram, S. (2023) · Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · doi:10.18653/v1/2023.acl-long.878 · Cited by 15
Arroyo-Machado, W., Ma, J., Chen, T., Johnson, T. P., Islam, S., Michalegko, L., & Welch, E. (2025) · PLOS One · doi:10.1371/journal.pone.0330416 · Cited by 15
Boissin, E., Costello, T. H., Spinoza-Martín, D., Rand, D. G., & Pennycook, G. (2025) · PNAS Nexus · doi:10.1093/pnasnexus/pgaf325 · Cited by 18
Towards Implicit Bias Detection and Mitigation in Multi-Agent LLM Interactions
Borah, A., & Mihalcea, R. (2024) · Findings of the Association for Computational Linguistics: EMNLP 2024 · doi:10.18653/v1/2024.findings-emnlp.545 · Cited by 15
MarIA and BETO are sexist: evaluating gender bias in large language models for Spanish
(2023) · doi:10.1007/s10579-023-09670-3 · Cited by 16
“Under these categories, we can see that there is a certain tendency to associate men with positive status in models such as BETO, MarIA, Geotrend, Amine Table 12 Differences between male and female for visibility-polarity categories or Recognai, and women with sentimental characteristics (love) in models such as MarIA…”
InfluentialHristova, T., Magee, L., & Soldatić, K. (2024) · AI & Society · doi:10.1007/s00146-024-02039-2 · Cited by 35
“…which often juxta-pose the production of utterances through statistical probability to true meaning and comprehension—famously through the critique of LLMs as stochastic parrots (Bender et al 2021)—our revisiting of the history of quantification of semantics suggests a more complicated picture.”
Munk, A. K., Jacomy, M., Ficozzi, M., & Jensen, T. E. (2024) · Big Data & Society · doi:10.1177/20539517241255107 · Cited by 25
“…have been the subject of a wide range of fruitful controversy analyses, all the way back to the early developments of neural networks in the 1950s (Olazaran, 1996), to recent debates about intentionality in large language models (Bender et al., 2021) or fairness in AI systems (Benbouzid, 2023).”
Krapp, E., Neuhaus, R., Hassenzahl, M., & Laschke, M. (2024) · Nordic Conference on Human-Computer Interaction · doi:10.1145/3679318.3685501 · Cited by 19
“LLMs, termed ”stochastic parrots” by critics [4], mimic patterns in their training data without genuine comprehension.”
LLM-based methods for the creation of unit tests in game development
Păduraru, C., Staicu, A.-N., & Ştefănescu, A. (2024) · Procedia Computer Science · doi:10.1016/j.procs.2024.09.473 · Cited by 19
Metaheuristics and Large Language Models Join Forces: Toward an Integrated Optimization Approach
Sartori, C. C., Blum, C., Bistaffa, F., & Corominas, G. R. (2024) · IEEE Access · doi:10.1109/access.2024.3524176 · Cited by 24
“In the influential paper by Bender et al. [48], the authors raise concerns about the risks of using LLMs and question their necessity.”
Resisting AI Solutionism: Where Do We Go From Here?
Reyes-Cruz, G., Spors, V., Muller, M. J., Felice, M. C., Bardzell, S., Williams, A., Hansson, K. E., & Feldfeber, I. (2025) · CHI Conference Extended Abstracts on Human Factors in Computing Systems (CHI EA) · doi:10.1145/3706599.3706732 · Cited by 21
“Misogyny, racism and other obscure views can come to the surface when large quantities of human data produced online is used as a basis in machine learning [6, 8, 46].”
The Role of ChatGPT in Dermatology Diagnostics
Khamaysi, Z., Awwad, M., Jiryis, B., Bathish, N., & Shapiro, J. (2025) · Diagnostics · doi:10.3390/diagnostics15121529 · Cited by 19
Doc2OracLL: Investigating the Impact of Documentation on LLM-Based Test Oracle Generation
Hossain, S. B., Taylor, R., & Dwyer, M. B. (2025) · Proceedings of the ACM on software engineering. · doi:10.1145/3729354 · Cited by 42
“For instance, fine-tuning smaller models generally requires fewer GPU resources, making them more cost-effective and reducing their environmental impact [4].”
Tschisgale, P., Maus, H., Kieser, F., Kroehs, B., Petersen, S., & Wulff, P. (2025) · Physical Review Physics Education Research · doi:10.1103/6fmx-bsnl · Cited by 37
“[96]), without considering whether such information should be treated as known or should be derived.”
The Impact of AI on Students’ Reading, Critical Thinking, and Problem-Solving Skills
Hassen, M. Z. (2025) · American Journal of Education and Information Technology · doi:10.11648/j.ajeit.20250902.12 · Cited by 28
Tres escenarios para la IA en educación: del apoyo responsable a la cocreación
García‐Peñalvo, F. J. (2025) · Education in the Knowledge Society (EKS) · doi:10.14201/eks.32932 · Cited by 19
Responsibility Gaps, LLMs & Organisations: Many Agents, Many Levels, and Many Interactions
Constantinescu, M., & Kaptein, M. (2025) · Science and Engineering Ethics · doi:10.1007/s11948-025-00560-1 · Cited by 23
“The risks of LLMs seem particularly noteworthy in terms of content toxicity (Dwivedi et al., 2023) and hallucinations (Bender et al., 2021) when used to generate advice in high-stakes domains, such as health, safety, legal or financial matters (Hagendorff, 2024).”
Jones, E. M., Newman, J. D., Kim, B., & Fogle, E. (2025) · JMIR Medical Education · doi:10.2196/80084 · Cited by 13
“Perhaps most significantly, this study does not address the enormous ethical concerns raised by slop, such as unmitigated biases, the unpaid and uncredited use of intellectual property for training, and the environmental impacts of needless AI use [41,50].”
Goyal, H., Garg, G., Mordia, P., Ramachandran, V., Kumar, D., & Challa, J. S. (2025) · Scientific Reports · doi:10.1038/s41598-025-18047-1 · Cited by 32
Bias in AI systems: integrating formal and socio-technical approaches
Ahmad, A., Vallès, Y., & Idaghdour, Y. (2026) · Frontiers in Big Data · doi:10.3389/fdata.2025.1686452 · Cited by 33
Does AI already have human-level intelligence? The evidence is clear
Chen, E. K., Belkin, M., Bergen, L., & Danks, D. J. (2026) · Nature · doi:10.1038/d41586-026-00285-6 · Cited by 26
A reporting checklist for large language models in behavioural science
Feuerriegel, S., Barrie, C., Crockett, M. J., Globig, L. K., McLoughlin, K. L., Mirea, D.-M., Spirling, A., Yang, D., Althoff, T., Antoniak, M., et al. (2026) · Nature Human Behaviour · doi:10.1038/s41562-026-02492-7 · Cited by 50
Peace and Prosperity for the Digital Age? The Colonial Political Economy of European AI Governance
Carmel, E., & Paul, R. (2022) · IEEE Technology and Society Magazine · doi:10.1109/mts.2022.3173340 · Cited by 25
Participatory Detection of Language Barriers towards Multilingual Sustainability(ies) in Africa
Litre, G., Hirsch, F., Caron, P., Andrason, A., Bonnardel, N., Fointiat, V., Nekoto, W., Abbott, J., Dobre, C., Rocha, J. D., et al. (2022) · Sustainability · doi:10.3390/su14138133 · Cited by 23
“Mobile technologies like Apple’s Siri, Google Assistant, and Amazon’s Alexa collectively use zero African languages [79], while the AI field continues to largely set aside multilingualism among Silicon Valley innovators [80].”
Fewer Errors, but More Stereotypes? The Effect of Model Size on Gender Bias
Tal, Y., Magar, I., & Schwartz, R. (2022) · Workshop on Gender Bias in Natural Language Processing (GeBNLP) · doi:10.18653/v1/2022.gebnlp-1.13 · Cited by 55
NeuroCounterfactuals: Beyond Minimal-Edit Counterfactuals for Richer Data Augmentation
Howard, P., Singer, G., Lal, V., Choi, Y., & Swayamdipta, S. (2022) · Findings of the Association for Computational Linguistics: EMNLP (Findings of EMNLP) · doi:10.18653/v1/2022.findings-emnlp.371 · Cited by 33
Unmasking Nationality Bias: A Study of Human Perception of Nationalities in AI-Generated Articles
Venkit, P. N., Gautam, S., Panchanadikar, R., Huang, T.-H., & Wilson, S. (2023) · Proceedings of the AAAI/ACM Conference on AI Ethics and Society · doi:10.1145/3600211.3604667 · Cited by 42
Generative AI poses ethical challenges for open science
Ación, L., Rajngewerc, M., Randall, G., & Etcheverry, L. (2023) · Nature Human Behaviour · doi:10.1038/s41562-023-01740-4 · Cited by 19
“Nonetheless, we know that data-based AI tools (including these new generative tools) can and do cause harm 2–4 .”
BackgroundAssessing deep learning: a work program for the humanities in the age of artificial intelligence
Segessenmann, J., Stadelmann, T., Davison, A., & Dürr, O. M. (2023) · AI and Ethics · doi:10.1007/s43681-023-00408-z · Cited by 24
Generative Information Systems Are Great If You Can Read
Roegiest, A., & Pinkosova, Z. (2024) · Conference on Human Information Interaction and Retrieval · doi:10.1145/3627508.3638345 · Cited by 22
Hardware Phi-1.5B: A Large Language Model Encodes Hardware Domain Specific Knowledge
Fu, W., Li, S., Zhao, Y., Ma, H., Dutta, R., Zhang, X., Yang, K., Jin, Y., & Guo, X. (2024) · Asia and South Pacific Design Automation Conference (ASP-DAC) · doi:10.1109/asp-dac58780.2024.10473927 · Cited by 30
“98 and an epsilon value of 1 e − 7 , was the chosen algorithm for its reliable performance in similar tasks.”
MethodChatbots, search engines, and the sealing of knowledges
Lindemann, N. F. (2024) · AI & Society · doi:10.1007/s00146-024-01944-w · Cited by 31
“They are voices which can be found e.g. on Wikipedia, which make up a large part of the content of most LLM training data and are likely to be included in the answers of chatbots as search engines (see Bender et al. 2021).”
We are Building Gods: AI as the Anthropomorphised Authority of the Past
Öhman, C. (2024) · Minds and Machines · doi:10.1007/s11023-024-09667-z · Cited by 35
“The answer is that Hägglund’s critique of God allows us to verbalize why even a perfect past, cured from the imperfections identified by the likes of Bender et al. (2021), Dressel and Farid, (2018), Diab, (2023) and O’Neal (2017), is an undesirable ruler.”
Evolving Generative AI: Entangling the Accountability Relationship
Elliott, M. T. J., Deepak, P., & MacCárthaigh, M. (2025) · Digital Government Research and Practice · doi:10.1145/3664823 · Cited by 30
Cahill, C., & McCabe, K. T. (2024) · PS Political Science & Politics · doi:10.1017/s1049096524000155 · Cited by 26
A foundation model enhanced approach for generative design in combinational creativity
Chen, L., Zhang, Y. E., Han, J., Sun, L., Childs, P., & Wang, B. (2024) · Journal of Engineering Design · doi:10.1080/09544828.2024.2356707 · Cited by 28
“Whilelargefoundationmodelshaveprovidedmomentum in the field of generative design, they currently lack the capability to create novel and ground-breaking design concepts (Bender et al. 2021).”
Animation and Artificial Intelligence
Stark, L. (2024) · ACM Conference on Fairness, Accountability, and Transparency (FAccT) · doi:10.1145/3630106.3658995 · Cited by 38
“Syntactically, ChatGPT’s representation of human text outputs are frequently accurate and coherent; however, they are intrinsically incapable of expressing semiotic meaning as made by humans using natural language [8].”
BackgroundVakili, T., Henriksson, A., & Dalianis, H. (2024) · BMC Medical Informatics and Decision Making · doi:10.1186/s12911-024-02546-8 · Cited by 24
“While PLMs have shown great promise in solving important NLP problems, their reliance on increasingly large numbers of parameters and vast corpora causes models to memorize parts of their training data [18–20].”
BackgroundApiola, M., Vartiainen, H., & Tedre, M. (2024) · ACM Conference on Innovation and Technology in Computer Science Education (ITiCSE) · doi:10.1145/3649217.3653596 · Cited by 23
Yang, W. (2024) · PLoS ONE · doi:10.1371/journal.pone.0306639 · Cited by 23
Oswal, S. K., & Oswal, H. K. (2024) · IEEE International Professional Communication Conference (ProComm) · doi:10.1109/procomm61427.2024.00030 · Cited by 21
Conversational AI and Chatbots: Enhancing User Experience on Websites
Dobbala, M. K., & Lingolu, M. S. S. (2024) · American Journal of Computer Science and Technology · doi:10.11648/j.ajcst.20240703.11 · Cited by 17
ChatGPT in the public eye: Ethical principles and generative concerns in social media discussions
Cohen, M., Khavkin, M., Davidow, D. M., & Toch, E. (2024) · New Media & Society · doi:10.1177/14614448241279034 · Cited by 27
“For instance, Bender et al. (2021) illustrated how large language models (LLMs) are susceptible to adopting and reflecting societal biases due to being trained on extensive and often uncurated Internet datasets.”
Uncovering labeler bias in machine learning annotation tasks
Haliburton, L., Leusmann, J., Welsch, R., Ghebremedhin, S., Isaakidis, P., Schmidt, A., & Mayer, S. (2024) · AI and Ethics · doi:10.1007/s43681-024-00572-w · Cited by 40
“However, datasets embed the cognitive biases of crowd workers [1], which can propagate through machine learning (ML) systems [2].”
BackgroundThe Language Model Revolution: LLM and SLM Analysis
Örpek, Z., Tural, B., & Destan, Z. (2024) · International Artificial Intelligence and Data Processing Symposium (IDAP) · doi:10.1109/idap64064.2024.10710677 · Cited by 29
He, S., & Lu, Y. (2024) · Interactive Learning Environments · doi:10.1080/10494820.2024.2415444 · Cited by 21
“Researchers must overcome these di ffi cul-ties, actively explore the new paradigm of integrating Gen AI and education (Bender et al., 2021), and create a local Gen AI suitable for the Chinese context to better enable Gen AI, empower education, and promote the digital transformation of education…”
A caution against customized AI in healthcare
Kostick, K. M. (2025) · npj Digital Medicine · doi:10.1038/s41746-024-01415-y · Cited by 18
Capano, G., He, A. J., & McMinn, S. (2025) · Journal of Asian Public Policy · doi:10.1080/17516234.2025.2450571 · Cited by 15
“For example, the potential for fabricated data or misattributed authorship undermines trust in scientific results (Bender et al., 2021).”
Pokhrel, S., Bina, K. C., & Shah, P. B. (2025) · Journal of Trends in Computer Science and Smart Technology · doi:10.36548/jtcsst.2024.4.007 · Cited by 21
Algorithmic-authors in academia: blurring the boundaries of human and machine knowledge production
Gretzky, M., & Dishon, G. (2025) · Learning Media and Technology · doi:10.1080/17439884.2025.2452196 · Cited by 21
“Additional concerns focus on the reliability and potential biases of AI-generated information, as LLMs may reproduce biases pre-sent in their training data or generate convincing but factually incorrect content (Bender et al. 2021; Williamson et al. 2023).”
Metadata
- Authors
- Bender, E. M., Gebru, T., McMillan-Major, A., & Shmitchell, S.
- Affiliations
- University of Washington (US)
- Black in AI, Palo Alto, CA, USA
- The Aether
- Venue
- Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency
- Publisher
- ACM
- Published
- 2021-03
- Pages
- 610-623
- OpenAlex
- W3133702157
- Cited by
- 6,988
- References
- 117
- Citations by year
- 2026: 2484 · 2025: 1887 · 2024: 1153 · 2023: 842 · 2022: 384 · 2021: 229
- FWCI
- 300.72
- Type
- conference paper
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- Locations
- ACM Conference on Fairness, Accountability, and Transparency (FAccT) (publishedVersion, cc-by)
- Creighton Digital Repository (CDR) (submittedVersion)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Topic Modeling
- Physical Sciences › Computer Science › Artificial Intelligence › Ethics and Social Impacts of AI
- Physical Sciences › Computer Science › Artificial Intelligence › Natural Language Processing Techniques
- Keywords
- large language models, model size, AI risks, financial costs, data curation, dataset documentation, stakeholder value, BERT, GPT, fine-tuning
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.