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 6,201–6,300 of 6,931
Amodio, C., & Diurni, A. (2025) · Springer eBooks · doi:10.1007/978-3-031-93247-2_1
Kim, M. C. (2026) · Religions · doi:10.3390/rel17080875 · Cited by 2
The Quality of Suicide-Related Stories Generated by Large Language Models: A Content Analysis
Sinyor, M., Chan, P. P. M., Men, V. Y., Danaee, E., Mason, S. G., Till, B., Robinson, J., Pirkis, J., Schaffer, A., Steinberg, R. J., et al. (2026) · Crisis · doi:10.1027/0227-5910/a001069
A survey of gender bias mitigation in neural machine translation
Gajakos, N., Staff, C., Murphy, B., Kelleher, J. D., & Haque, R. (2026) · Language Resources and Evaluation · doi:10.1007/s10579-026-09931-x
“This bias often arises from historical imbalances or stereotypes present in the training data (Blodgett et al., 2020; Bender et al., 2021).”
Fokoué, E. (2026) · Sustainability · doi:10.3390/su18157545 · Cited by 2
Capability determinism, energy and AI labour substitution
Mbioh, W. R. (2026) · AI & Society · doi:10.1007/s00146-026-03251-y
Why parents of children aged 4–6 use generative AI: a qualitative interview study
Güngör, B. Ş., & Güngör, A. (2026) · Children and Youth Services Review · doi:10.1016/j.childyouth.2026.109196
Truth-ε and machine knowledge: from demonstrative rationality to bounded computational convergence
Ikonomopoulos, S. (2026) · Proceedings of the European Academy of Sciences and Arts · doi:10.4081/peasa.84
“It depends on calibration, verification, robustness, and error con-trol (Bender et al., 2021).”
Yazıcı, A. M. (2026) · Systems Research and Behavioral Science · doi:10.1002/sres.70116
Safaei, D., Hein, A., & Haki, K. (2026) · Journal of Information Technology · doi:10.1177/02683962261467413 · Cited by 2
“Unlike traditional models with fixed input–output mappings, GenAI models generate outputs probabilistically, meaning identical inputs can yield different outputs across executions (Bender et al., 2021; Brown et al., 2020).”
Should sensitive archival images be shown? The case for applying AI to colonial archives
Jaillant, L., Kidd, M., & Zhao, L. (2026) · Archival Science · doi:10.1007/s10502-026-09552-x · Cited by 1
“Bender et al. (2021) agree, suggesting that large-scale language and vision models trained on uncurated internet data “encode the dominant/hegemonic view, which further harms people at the margins,” since “the voices of people most likely to hew to a hegemonic viewpoint are also more likely to be…”
AI for Financial Investment Decision Making
Romprasert, S., Romprasert, T., & Tanamee, D. (2026) · Communications in computer and information science · doi:10.1007/978-3-032-33384-1_4
When AI speaks: user imaginaries of generative AI and freedom of expression
Brandtzæg, P. B., & Wang, X. (2026) · AI & Society · doi:10.1007/s00146-026-03188-2
Looking for Shakespeare: The Bard in Big Data
Rodriguez, A. (2026) · Shakespeare · doi:10.1080/17450918.2026.2706475
Liu, S., Duan, H., Zhou, Y., Zhao, H., Tan, X., & Li, H. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.7180408
Gender-Responsive Generative AI–Empowered Entrepreneurship: Challenges and Opportunities
Nikghadam-Hojjati, S., Barata, J., & Marchetti, E. (2026) · IntechOpen eBooks · doi:10.5772/intechopen.1017036 · Cited by 1
Exploring the potential of generative AI and teacher collaboration in evaluating academic papers
Zhong, Q., Ji, H., Liu, T., Li, F., & Wang, X. (2026) · Education and Information Technologies · doi:10.1007/s10639-025-13864-3
Literacidad en IA: Funcionamiento, Sesgos y Operatividad de los LLMs
Joaquín, R. A. R. (2026) · Espila · doi:10.61454/yhetdk61
Pant, Y. R., Adarbah, H. Y., Noore, A., Che, D., Ahmed, A., & Ayala, R. (2026) · AI · doi:10.3390/ai7080280
Rokaya, M., AlZain, M. A., Hemdan, D. I. I., Atwa, M. A., & Atlam, E.-S. (2026) · Journal of Disability Research · doi:10.57197/jdr-2026-0888 · Cited by 1
Do people rely on ChatGPT more than their peers to detect deepfake news?
Fu, Y., & Hanaki, N. (2026) · Journal of Economic Interaction and Coordination · doi:10.1007/s11403-026-00490-6 · Cited by 1
Durango, R. A. B., Correa, D. A. C., Flores, E. A. R., & Zuluaga, C. D. J. G. (2026) · Philosophies · doi:10.3390/philosophies11040133
“This temporal pattern, cap‑ tured in Figure 3, supports treating AI‑related responsibility discourse as a rapidly expand‑ ing, cross‑cutting research front rather than a marginal subtopic [15,31,32,85].”
Steinhoff, J. (2026) · Oxford University Press eBooks · doi:10.1093/9780197851395.003.1536
What We Govern as Intelligence: Artificial Intelligence and The Problem of Answerability
Tomičić, A., & Gjorgjioska, M. A. (2026) · Digital Society · doi:10.1007/s44206-026-00274-2
“In this respect, critiques of large language models as systems optimized for scale, pat - tern continuation, and plausible output remain highly relevant (Bender et al., 2021 ).”
Puri, S., Dubey, R., Pandey, S., & Saha, R. (2026) · Innovations in Education & Teaching International · doi:10.1080/14703297.2026.2709762
Galasso, B. (2026) · Springer eBooks · doi:10.1007/60490_2026_32
Harada, Y. (2026) · Cureus Journal of AI-Augmented Research · doi:10.7759/s44568-026-00185-3
Bilici, M. V. (2026) · İstanbul Üniversitesi Sosyoloji Dergisi / İstanbul University Journal of Sociology · doi:10.26650/sj.2025.46.1.1172
Grasnick, A. (2026) · Springer eBooks · doi:10.1007/978-3-658-48558-0_5
Almazroui, K. M. (2026) · Education Sciences · doi:10.3390/educsci16081202
Toward a phenomenology of mystification: The ambivalence of magic among AI technologists
Ochigame, R., & Jones, G. M. (2026) · Anthropological Theory · doi:10.1177/14634996261470761 · Cited by 1
“The best-known critiques of GenAI have worked largely in the register of demysti fi - cation, for example by describing LLMs as “ stochastic parrots ” (Bender et al., 2021; Mitchell and Krakauer, 2023) that can, at best, only mimic human language.”
Three Stance Corridors: Hedging, Engagement, and Task-Flex in AI-Generated Academic WritingPreprint
Curfman, E. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.7195350
Made in the Image of AI: Evangelicalism and Machine-Generated Biblical Media
Zeser, K. (2026) · Open Theology · doi:10.1515/opth-2025-0076
Building Grounded Synthetic Users: Generative AI as Mediating Artefact in User Research
Zhong, K., Figoli, F. A., & Rampino, L. (2026) · International Conference on Human-AI Interaction and Experience Design (HAXD) · doi:10.1109/haxd70072.2026.11620952
“These observations reinforce the importance of ongoing human interpretation and qualitative oversight in AI-mediated user research [31], [32].”
Human-in-the-Loop Learning for Improving Language Model Reliability
Dandigam, R., Patel, M., Pareek, C. S., Faysal, S. M., Alam, G. T., & Siam, M. A. (2026) · International Conference on Intelligent Technologies (CONIT) · doi:10.1109/conit69683.2026.11620666 · Cited by 2
“Additionally, fully automated learning mechanisms may not have human reasoning, ethical judgement, and domain-specific contextual validation, leading to decreased robustness and interpretability [5].”
Thinking Fast and Planning Slow: Language-Guided Latent Control with Hierarchical Value Functions
Ansah, H. F., Saneei, S., & Banerjee, S. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-33010-9_1
Incorporation of generative AI into science communication education in multilingual environments
Qing, H. C., Melo, J. D., Freedman, L., & Andreopoulos, S. (2026) · Discover Education · doi:10.1007/s44217-026-01891-2
“Overrepresentation of Western, male, and English-centric content in the training data introduces latent systemic biases in model outputs, raising important ethical concerns within the academic community regarding large-scale AI models [46, 47].”
S-AI-Reasoning: Reasoning as Regulated CanonicalizationPreprint
Slaoui, S. C. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.6888738 · Cited by 5
Comes, L., Silveira, J. C. S., Martins, H. C., Lana, C. A., & Villela, M. L. B. (2026) · Journal on Interactive Systems · doi:10.5753/jis.2026.7388 · Cited by 1
Choudhary, K. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.7013978
Reina, L. A. V., Ruiz-Fernández, V., & Falcão, J. (2026) · Compàs d amalgama · doi:10.1344/compas2026.13.8
Truth According to Machine: AI’s Influence on Human Credibility AssessmentPreprint
Glowacki, Z. R., Raffloer, G., & Frank, M. G. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.7202488
Xiao, H. (2026) · International Conference on Big Data & Artificial Intelligence & Software Engineering (ICBASE) · doi:10.1109/icbase70763.2026.11619439
“[11] To address this issue, many current LLMs have defensive mechanisms to detect and prevent generating biased, discriminatory and other kinds of inappropriate content.”
Slaoui, S. C. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.6818118
Chen, W. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.7017078
Seeing against the machine: reframing generative AI error as pedagogical insight in art education
Abdulmajid, M. (2026) · Journal of Visual Literacy · doi:10.1080/1051144x.2026.2710031
Psicoterapia e Intelligenza Artificiale: possibilità o minaccia?
Resicato, G. (2026) · RIVISTA DI PSICOTERAPIA RELAZIONALE · doi:10.3280/pr2026-063002
Leyme-Pilicita, K. E., Paucar-Jiménez, L. F., & Zambrano, M. F. G. (2026) · Innova Science Journal · doi:10.63618/omd/isj/v4/n3/390 · Cited by 1
“…por los estudiantes, asimismo, a esto se suman los sesgos algorítmicos que pueden condicionar la interpretación de la información junto con la construcción argumentativa, al reproducir perspectivas parciales, hegemónicas o culturalmente limitadas (Bender et al., 2021; Miao & Holmes, 2023).”
Li, C., & Zhang, M. (2026) · Annals of the New York Academy of Sciences · doi:10.1111/nyas.70360
Valdivieso-Mora, M. C., Intriago-Intriago, E. A., Faubla-Vera, L. A., & Chávez-Moreira, A. N. (2026) · Innova Science Journal · doi:10.63618/omd/isj/v4/n3/334 · Cited by 1
Meeting the machines half-way: Testing LLMs for qualitative coding of values in texts
Moats, D., & Žagar, A. P. (2026) · Human Organization · doi:10.1080/00187259.2026.2703871 · Cited by 1
Hemant, P., & Arif, A. (2026) · Journal of Grid Computing · doi:10.1007/s10723-026-09844-7
Firmino, V. P., Lopes, J. N. D. S., dos Reis, V. Q., de Lima, A. C., & Nogueira, B. M. (2026) · Anais do Workshop sobre Educação em Computação (WEI) · doi:10.5753/wei.2026.22017
Rethinking Anticipation in the Age of Generative AI: Toward a Theory of Hybrid AnticipationPreprint
Lee, T. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.7212118
Rizzo, A. ”., Mozgai, S., Hale, B., & Shilling, R. D. (2026) · Elsevier eBooks · doi:10.1016/b978-0-443-33151-0.00003-4 · Cited by 2
Processing versus nutrients: What drives the health risks of ultra-processed foods?Preprint
Sun, Q., Wang, Y., Fan, Z., Zhu, N., Mu, K., Nie, Q., Zheng, J., Gao, C., Xiao, Y., Yang, X., et al. (2026) · Research Square · doi:10.21203/rs.3.rs-10513803/v1
From Accommodation to Anticipation: Using GenAI and UDL to Design Accessible TEFL Lessons
Saykılı, A. (2026) · IGI Global eBooks · doi:10.4018/979-8-3373-8460-3.ch008 · Cited by 1
Automating Observational Coding of Psychotherapy Quality with Natural Language Processing (Preprint)Preprint
Benito, K., Jangraw, D. C., Kemp, J. J., Barker, D., Han, E., Norris, L. A., Danforth, C. M., Smits, J. A. J., Otto, M. W., Freeman, J. B., et al. (2026) · doi:10.2196/preprints.107147
Akbarian, S. (2026) · IGI Global eBooks · doi:10.4018/979-8-3373-8460-3.ch006
Bao, P. (2026) · Current Issues in Language Planning · doi:10.1080/14664208.2026.2711654 · Cited by 2
Dede, O., Ataseven, H., & Bökeoğlu, Ö. Ç. (2026) · IGI Global eBooks · doi:10.4018/979-8-3373-8460-3.ch002
Betancourt, A., COYAGO, G. A. A., Sangoluisa, K., Auz-Torres, M., & Yoo, S. G. (2026) · IEEE Access · doi:10.1109/access.2026.3718389
“…significant number of references in literature at 79 instances, was the difficulty in verifying the correctness of AI-generated responses by analysts, this issue is related to ‘‘the tendency of LLMs to produce hallucinations, where the model generates plausible but incorrect information’’ [24].”
Generative AI in curriculum design: Advancing or undermining decolonization efforts?
Dawarka, V., & Huzooree, G. (2026) · Elsevier eBooks · doi:10.1016/b978-0-443-36775-5.00005-9
Peirce's revenge on the Chinese Room
Loginov, E., Vasilyev, V. V., Kuznetsov, A. V., Volkov, D. B., Беседин, А. П., & Mertsalov, A. V. (2026) · Frontiers in Psychology · doi:10.3389/fpsyg.2026.1900073 · Cited by 1
Schmidt, R. (2026) · Learning Media and Technology · doi:10.1080/17439884.2026.2708693
Cross-Linguistic Epistemic Variance in Large Language Model Cognition
Nizami, M. S., Khan, M. R. S., & Bilal¹, M. (2026) · International Journal of Innovations in Science and Technology · doi:10.33411/ijist/1932 · Cited by 1
The Creative Process in AI-Driven Marketing
Stiefenhofer, P. (2026) · Springer eBooks · doi:10.1007/978-3-032-22336-4_7
Relecturas de la tradición semio-discursiva para el estudio de la comunicación digital
Cárdenas, T. K. (2026) · Global Media Journal México · doi:10.29105/gmjmx23.45-576
Qleibo, A. (2026) · Journal of Palestine Ahliya University for Research and Studies · doi:10.59994/pau.2026.2.71
Fewer Clicks Matter: AI’s Ability to Improve Website Usage for Educator Professional Development
Rowland, A. L., Mosher, M. A., Goldman, S. R., Smith, S. J., & Carreon, A. (2026) · Journal of Special Education Technology · doi:10.1177/01626434261476316
Khatami, M. (2026) · Subjectivity · doi:10.1057/s41286-026-00244-6
“…non-Western philosophical thought, and manifestos of emergent social movements are either profoundly underrepresented in these corpora or, if present, are often filtered out in subsequent layers of corporate-engineered “safety filtering” and “harmful content” removal (Bender et al. 2021).”
Sentinel Polity: LLM-Grounded Policy-as-Code for IaC Hardening in Hyperscale Clouds
Amin, F. (2026) · Lecture notes in networks and systems · doi:10.1007/978-3-032-24360-7_30
New Accelerants of Inequality Regimes: AI‐Hiring Tools and Algorithmic Bias
Yarrow, E. (2026) · Gender Work and Organization · doi:10.1111/gwao.70232 · Cited by 2
Optimizing Artificial Intelligence for Sustainability: Computational Efficiency and Green Metrics
Das, S., Singh, S., Garg, A., Gaur, M. P., Pandita, S., & Verma, N. (2026) · Lecture notes in networks and systems · doi:10.1007/978-3-032-28594-2_7
Dangerous Bullshit: Why It Is Wrong to Hold Clinicians Responsible for the Outputs of Generative AI
Daus, Z., Howard, M., & Sparrow, R. (2026) · The American Journal of Bioethics · doi:10.1080/15265161.2026.2690954
“In this sense, LLMs are not designed to produce text that accurately represents the world, but text that convincingly “parrots” the corpuses that are included in the datasets on which they are trained (Bender et al. 2021).”
Cross-LLM Paraphrase Laundering: A Register-Controlled Evaluation of Fake News Detectors
Aliguliyev, R. M., & Mehdiyev, J. (2026) · AI · doi:10.3390/ai7080297 · Cited by 1
SynergyScale: Optimizing Offloading and Task Partitioning for Efficient Model Training
Sun, X., Xu, J., & Wang, Z. (2026) · IEEE Transactions on Parallel and Distributed Systems · doi:10.1109/tpds.2026.3719351
“Distributed training methods, including tensor [13], [14] and pipeline parallelism [15], [11], often suffer from low hardware utilization, as GPUs and CPU cores remain idle during synchronization, pipeline bubbles, or data stalls [20], [21].”
Luiz, A. A. O. D., Amaral, P. A. V. D., & Neri, J. B. (2026) · Revista de Geopolítica · doi:10.56238/revgeov17n8-010
Does Expertise Still Matter? Generative AI and the Crisis of Professional Identity
Nyagoga, L. M. (2026) · Global Journal of Human-Social Science · doi:10.34257/gjhssg258994
Bag-of-words or embedding-based topic models? Comparing LDA, BERTopic, and six other methods
Lamirel, J.-C., Lareau, F., Prouteau, T., & Malaterre, C. (2026) · Scientometrics · doi:10.1007/s11192-026-05745-4 · Cited by 3
“First, LLM-based topic modelling frequently depends on proprietary systems whose training data, model architecture, and internal reasoning processes are opaque, raising issues around transparency and accountability (Bender et al., 2021).”
Hefter, M. H., Paaßen, B., & Berthold, K. (2026) · AI in Education · doi:10.3390/aieduc2030027 · Cited by 1
“Such knowledge should include, for instance, that a genAI system is a statistical text-completion machine rather than “intelligent” in a human sense (Bender et al., 2021; Ilievski et al., 2025).”
Singh, S., Chandra, M., & Awasthi, D. (2026) · Lecture notes in networks and systems · doi:10.1007/978-3-032-28594-2_45
AI Agents, Data Power, and the Ideological Horizon of AGI
Gibson, Q. H. (2026) · International Journal of Human-Computer Interaction · doi:10.1080/10447318.2026.2710519
Assheton, G., Kioupi, V., & Michel‐Villarreal, R. (2026) · World sustainability series · doi:10.1007/978-3-032-22076-9_38
Hamidavi, A. (2026) · Research Square · doi:10.21203/rs.3.rs-10560655/v1 · Cited by 1
Análise de Falhas na Detecção de Alucinação em Textos Jurídico-Policiais em Português
Barcelar, R. R., & Ventura, T. M. (2026) · doi:10.5753/pesquisanuvem.2026.23945
“Conforme discutido por Bender et al. (2021), modelos de linguagem operam por predição estatística sem compromisso intrínseco com a veracidade.”
Richardt, C. D. C. L. T., Martins, N. P., Galafassi, C., & do Amaral, É. M. H. (2026) · Anais do Workshop sobre as Implicações da Computação na Sociedade (WICS) · doi:10.5753/wics.2026.23462
“Esses aspectos dialogam diretamente com a literatura recente sobre integridade acadêmica, vieses algorítmicos e responsabilização institucional na adoção de sistemas baseados em inteligência artificial [Bender et al. 2021; Dwivedi et al. 2023].”
Subero-Sáenz, Á. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.7231378
Hybrid LLM-Based Framework for Automated Subjective and MCQ Evaluation
Goel, A., Benjamen, I., Reddy, S. V. A., et al. (2026) · doi:10.1109/icdca69396.2026.11620055
The Non-Fakeness of Fake News: when alleged misinformation turns out to be true
Badawy, W. (2026) · AI and Ethics · doi:10.1007/s43681-026-01289-8
Jimenez, P., Kim, Y., & Fitzgerald, R. (2026) · AI & Society · doi:10.1007/s00146-026-03164-w
“Bender et al. (2021) argue these models can be seen as “stochastic parrots”—systems that produce fluent language output by probabilistically stringing together word sequences without any genuine grasp of meaning or communicative intent.”
Agentic AI Safety: A Structured Review of Open Problems and Their Regulatory Anchoring
Valenta, T., Rozinek, O., & Horálek, J. (2026) · AI · doi:10.3390/ai7080298 · Cited by 2
Carvalho, D. B. F., Souza, E. D. P., Dias, B. P. M., Detomi, G. H. A., Sousa, P. V. F., Dias, W. M., & Viana, M. C. (2026) · Anais do Workshop sobre as Implicações da Computação na Sociedade (WICS) · doi:10.5753/wics.2026.21040
“A complexidade desse problema é agravada pelo fato de que LLMs, ao produzirem respostas linguisticamente coerentes e assertivas, tendem a ser percebidos como autoridades epistêmicas (Bender et al. , 2021; Weidinger et al. , 2021).”
Peters, M. A., & Yang, Y. (2026) · Educational Philosophy and Theory · doi:10.1080/00131857.2026.2706550
Gutiérrez, J. L. M., Hipolito, B., & Romero, D. (2026) · AI and Ethics · doi:10.1007/s43681-026-01255-4
A Critical Discourse Analysis (CDA) of ChatGPT’s responses to English grammar learning queries
Salaeh, A., Maming, A., Haeweasani, H., Sebastian, J. M., Pontongadil, A. D. P., Malfiya, T., Munida, V. F., & Puspitasari, D. (2026) · Priviet Social Sciences Journal · doi:10.55942/pssj.v6i8.2088
“Rather than producing neutral or purely objective outputs, algorithmic systems generate responses shaped by large-scale training data that reflect historical inequalities, cultural biases, and dominant linguistic hierarchies (Bender et al., 2021; Noble, 2018).”
Dougherty, J. P., & Gallenstein, R. (2026) · AI & Society · doi:10.1007/s00146-026-03240-1
Teacher Mediation and the Contingent Promise of Generative AI in Mathematics Education
Canonigo, A. M. (2026) · Journal of Computer Assisted Learning · doi:10.1002/jcal.70307 · Cited by 2
Gendered asymmetries in AI stories in English and Spanish
Wentworth, I. (2026) · AI & Society · doi:10.1007/s00146-026-03163-x
“Emily Bender et al., and many others since, have noted “large uncurated, Internet-based datasets encode the dominant/hegemonic view, which further harms people at the margins” (Bender et al. 2021, p. 613).”
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.