CoChat
Sign Up
Cited by 180Ethics and Social Impacts of AITopic ModelingHate Speech and Cyberbullying Detection

Detecting Emergent Intersectional Biases: Contextualized Word Embeddings Contain a Distribution of Human-like Biases

Open access
Guo, Wei, Caliskan, Aylin·2021·Proceedings of the 2021 AAAI/ACM Conference on AI, Ethics, and Society
DOI 10.1145/3461702.3462536
Open this paper in CoChat — free accountOpen PDF

Abstract

With the starting point that implicit human biases are reflected in the statistical regularities of language, it is possible to measure biases in English static word embeddings. State-of-the-art neural language models generate dynamic word embeddings dependent on the context in which the word appears. Current methods measure pre-defined social and intersectional biases that occur in contexts defined by sentence templates. Dispensing with templates, we introduce the Contextualized Embedding Association Test (CEAT), that can summarize the magnitude of overall bias in neural language models by incorporating a random-effects model. Experiments on social and intersectional biases show that CEAT finds evidence of all tested biases and provides comprehensive information on the variance of effect magnitudes of the same bias in different contexts. All the models trained on English corpora that we study contain biased representations. GPT-2 contains the smallest magnitude of overall bias followed by GPT, BERT, and then ELMo, negatively correlating with the contextualization levels of the models.

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1145/3461702.3462536

Cites

39 of 39 references verified0 citation errors · 0 not checkable

Statistics

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

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?

  • Much of existing literature on diversity in AI focuses on demographic, cultural, or political group representation.

    “Much of the existing literature on diversity in AI asks whether different demographic, cultural, or political groups are represented or treated fairly (Hardt et al., 2016;Barocas et al., 2023;Guo & Caliskan, 2021;Wang et al., 2025).”

    From On Epistemic Diversity in Large Language Models (2026) · cites this paper as Guo & Caliskan, 2021

    The Introduction (page not specified) says most work on bias in word embeddings focuses on single social categories such as gender or race. That supports the demographic-group fairness aspect, but the paper does not establish the broader claim about diversity literature across AI or its focus on cultural and political group representation. (LLM, conf=0.83)

    Experimental — accuracy not yet measured

  • Meta-learning centers on equipping AI models with ability to rapidly adapt to new tasks and domains using limited data samples.

    “Meta-learning, or 'learning to learn', centers on equipping AI models with the ability to rapidly adapt to new tasks and domains using limited data samples [251], [252].”

    From From Google Gemini to OpenAI Q* (Q-Star): A Survey on Reshaping the Generative Artificial Intelligence (AI) Research Landscape (2025) · cites this paper as [252]

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Cites

Flagged first, then by citations

Research works printed but not linked: 26

  1. Su Lin Blodgett, Solon Barocas, Hal Daumé III, and Hanna Wallach. 2020. Lan- guage (Technology) is Power: A Critical Survey of" Bias" in NLP. arXiv preprint arXiv:2005.14050 (2020). · arXiv
  2. Bernd Bohnet, Ryan McDonald, Goncalo Simoes, Daniel Andor, Emily Pitler, and Joshua Maynez. 2018. Morphosyntactic tagging with a meta-bilstm model over context sensitive token encodings. arXiv preprint arXiv:1805.08237 (2018). · arXiv
  3. Tolga Bolukbasi, Kai-Wei Chang, James Y Zou, Venkatesh Saligrama, and Adam T Kalai. 2016. Man is to computer programmer as woman is to homemaker? debiasing word embeddings. In Advances in neural information processing systems. 4349-4357.
  4. Joy Buolamwini and Timnit Gebru. 2018. Gender shades: Intersectional accu- racy disparities in commercial gender classification. In Conference on fairness, accountability and transparency. 77-91.
  5. Aylin Caliskan, Joanna J Bryson, and Arvind Narayanan. 2017. Semantics derived automatically from language corpora contain human-like biases. Science 356, 6334 (2017), 183-186.
  6. Ciprian Chelba, Tomas Mikolov, Mike Schuster, Qi Ge, Thorsten Brants, Phillipp Koehn, and Tony Robinson. 2013. One billion word benchmark for measuring progress in statistical language modeling. arXiv preprint arXiv:1312.3005 (2013). · arXiv
  7. Rebecca DerSimonian and Nan Laird. 1986. Meta-analysis in clinical trials. Controlled clinical trials 7, 3 (1986), 177-188.
  8. Anthony G Greenwald and Mahzarin R Banaji. 1995. Implicit social cognition: attitudes, self-esteem, and stereotypes. Psychological review 102, 1 (1995), 4.
  9. Ange-Marie Hancock. 2007. When multiplication doesn't equal quick addition: Examining intersectionality as a research paradigm. Perspectives on politics 5, 1 (2007), 63-79.
  10. Rachel T Hare-Mustin and Jeanne Marecek. 1988. The meaning of difference: Gender theory, postmodernism, and psychology. American psychologist 43, 6 (1988), 455.
  11. Vasileios Hatzivassiloglou and Kathleen McKeown. 1997. Predicting the semantic orientation of adjectives. In 35th annual meeting of the association for computa- tional linguistics and 8th conference of the european chapter of the association for computational linguistics. 174-181.
  12. Larry V Hedges and Ingram Olkin. 2014. Statistical methods for meta-analysis. Academic press.
  13. Sepp Hochreiter and Jürgen Schmidhuber. 1997. Long short-term memory. Neural computation 9, 8 (1997), 1735-1780.
  14. Keita Kurita, Nidhi Vyas, Ayush Pareek, Alan W Black, and Yulia Tsvetkov. 2019. Quantifying Social Biases in Contextual Word Representations. In 1st ACL Workshop on Gender Bias for Natural Language Processing.
  15. Chandler May, Alex Wang, Shikha Bordia, Samuel R Bowman, and Rachel Rudinger. 2019. On measuring social biases in sentence encoders. arXiv preprint arXiv:1903.10561 (2019). · arXiv
  16. Tomas Mikolov, Ilya Sutskever, Kai Chen, Greg S Corrado, and Jeff Dean. 2013. Distributed representations of words and phrases and their compositionality. In Advances in neural information processing systems. 3111-3119.
  17. Moin Nadeem, Anna Bethke, and Siva Reddy. 2020. StereoSet: Measuring stereo- typical bias in pretrained language models. arXiv preprint arXiv:2004.09456 (2020). · arXiv
  18. Matthew E Peters, Mark Neumann, Mohit Iyyer, Matt Gardner, Christopher Clark, Kenton Lee, and Luke Zettlemoyer. 2018. Deep contextualized word representations. arXiv preprint arXiv:1802.05365 (2018). · arXiv
  19. 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.
  20. Emma Strubell, Ananya Ganesh, and Andrew McCallum. 2019. Energy and policy considerations for deep learning in NLP. arXiv preprint arXiv:1906.02243 (2019). · arXiv
  21. Yi Chern Tan and L Elisa Celis. 2019. Assessing social and intersectional biases in contextualized word representations. In Advances in Neural Information Processing Systems. 13209-13220.
  22. Autumn Toney and Aylin Caliskan. 2020. ValNorm: A New Word Embedding Intrinsic Evaluation Method Reveals Valence Biases are Consistent Across Lan- guages and Over Decades. arXiv preprint arXiv:2006.03950 (2020). · arXiv
  23. Autumn Toney, Akshat Pandey, Wei Guo, David Broniatowski, and Aylin Caliskan. 2020. Pro-Russian Biases in Anti-Chinese Tweets about the Novel Coronavirus. arXiv preprint arXiv:2004.08726 (2020). · arXiv
  24. Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. 2017. Attention is all you need. In Advances in neural information processing systems. 5998-6008.
  25. Rob Voigt, David Jurgens, Vinodkumar Prabhakaran, Dan Jurafsky, and Yulia Tsvetkov. 2018. RtGender: A corpus for studying differential responses to gender. In Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC 2018).
  26. 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. 5754-5764.

Cited by

Sorted by citations

Related

Authors
Guo, W., & Caliskan, A.
Affiliations
George Washington University (US)
Venue
Proceedings of the 2021 AAAI/ACM Conference on AI, Ethics, and Society
Publisher
ACM
Published
2021-07-21
Pages
122-133
ISSN
3065-8365 (ISSN-L)
DOI
10.1145/3461702.3462536
OpenAlex
W3034115845
Cited by
180
References
26
Citations by year
2026: 16 · 2025: 27 · 2024: 34 · 2023: 44 · 2022: 45 · 2021: 11 · 2020: 3
FWCI
30.34
Type
conference paper
Language
en
Open access
Licence
Locations
  • Proceedings of the AAAI/ACM Conference on AI Ethics and Society (publishedVersion, cc-by-nc-sa)
  • arXiv (Cornell University) (submittedVersion)
  • arXiv (Cornell University)
Topics
  • Social Sciences › Safety Research › Ethics and Social Impacts of AI
  • Physical Sciences › Computer Science › Artificial Intelligence › Topic Modeling
  • Physical Sciences › Computer Science › Artificial Intelligence › Hate Speech and Cyberbullying Detection
Keywords
intersectional bias, contextual word embeddings, social bias, language models, GPT-2, BERT, ELMo, random effects models, bias measurement, English language

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