Pith. sign in

REVIEW 4 major objections 5 minor 2 cited by

When Algorithms Play Favorites: Lookism in the Generation and Perception of Faces

T0 review · 4 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Text-to-image models link facial attractiveness to unrelated positive traits, and gender classifiers misclassify faces generated with negative trait labels more often, with the largest effects for non-White women.

desk verdict A promising but overreaching workshop study: the prompt-conditioned association between attractiveness words and trait words in SD2.1 output is real, but 'lookism' is not established without human ratings or an independent embedding. read the letter →

arxiv 2506.11025 v1 pith:O5GEG2PY submitted 2025-05-20 cs.LG cs.AIcs.CV

Miriam Doh , Aditya Gulati , Matei Mancas , Nuria Oliver This is my paper · ORCID
classification cs.LGcs.AIcs.CV
keywords facesalgorithmsclassificationgendergeneratedlookismsyntheticallysystems
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This is a measurement study, not a new algorithm. The authors generated 13,200 faces with Stable Diffusion 2.1 by varying gender, race, and five pairs of words: attractive or unattractive, intelligent or unintelligent, trustworthy or untrustworthy, sociable or unsociable, and happy or unhappy. For each group they compared the average CLIP image embedding, a numerical summary of visual content.

The first main result is that faces generated with positive words were, on average, closer in embedding space to faces generated with 'attractive', while faces generated with negative words were closer to 'unattractive'. The pattern was strongest for Asian and Black women and weaker or inconsistent for White faces. This suggests the image generator produces similar visual features for attractiveness and for unrelated positive traits.

The second main result is that three gender classifiers, InsightFace, DeepFace, and FairFace, were run on all images. Accuracy was high for male faces and for attractive female faces, but dropped sharply for female faces generated with negative words. For example, DeepFace correctly classified only about 12 percent of unhappy women and 20 percent of unsociable women, while FairFace stayed above 90 percent. The authors interpret this as algorithmic lookism: female faces not marked as attractive are harder to classify.

A key caveat is that the paper does not measure attractiveness directly. It uses the words in the prompts and the similarity of computer vision embeddings, and it lists as future work separating the effect from biases in the CLIP model itself.

Extended reading notes

Core claim

The central claim is that text-to-image systems exhibit algorithmic lookism: faces generated with positive trait descriptors are closer in CLIP embedding space to faces generated with 'attractive', faces generated with negative descriptors are closer to 'unattractive', and gender classification models show higher error rates on these 'less-attractive' faces, with Asian and Black women disproportionately affected. The abstract states: 'text-to-image (T2I) systems tend to associate facial attractiveness to unrelated positive traits like intelligence and trustworthiness; and (2) gender classification models exhibit higher error rates on less-attractive faces, especially among non-White women.'

Load-bearing premise

The measurement chain assumes that words in the prompt produce faces that humans would agree carry the trait, and that Euclidean distance between CLIP embedding centroids is a valid proxy for perceived trait association. This is load-bearing because both the T2I association result and the 'less-attractive' label in the gender classification result depend on it. The paper states in Section 4 that it 'does not define or measure attractiveness', and lists CLIP-bias disentanglement only as future work (Section 4, item 1). If prompts like 'unhappy' shift expression, makeup, or age rather than attractiveness, the classifier error pattern may reflect those visual confounds rather than lookism.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The paper reports an experimental study of algorithmic lookism in Stable Diffusion 2.1. The authors generate 13,200 face images varying gender, race, and five trait pairs (attractive/unattractive, intelligent/unintelligent, trustworthy/untrustworthy, sociable/unsociable, happy/unhappy), compute CLIP embedding centroids per group, and measure Euclidean distances between positive/negative trait groups and the attractive/unattractive groups. They also evaluate three gender classifiers (InsightFace, DeepFace, FairFace) on the same images. The two central claims are: (1) T2I models associate facial attractiveness with unrelated positive traits and unattractiveness with negative traits; and (2) gender classifiers have higher error rates on 'less-attractive' faces, especially Asian and Black women.

Significance. If the operationalization were validated, this would be a useful, compact demonstration of lookism in a current T2I model and its downstream effect on gender classifiers. The study has clear strengths: no model parameters are fitted, the evaluation uses three external classifiers, the prompt design is simple and reproducible, and the authors are transparent about several limitations. However, the two headline claims rest on an unvalidated mapping from prompt adjectives to perceived facial attributes, and the first claim additionally uses CLIP embeddings from the same model family that conditions the generator. These issues are load-bearing rather than cosmetic, so the current evidence supports a conditional finding rather than a definitive one.

major comments (4)
  1. [Section 2, similarity score definition; Section 4, item (1)] The first headline claim ('T2I systems tend to associate facial attractiveness with unrelated positive traits') is vulnerable to circularity because Stable Diffusion 2.1 is conditioned with a CLIP text encoder and the evaluation uses CLIP image embeddings. Although no parameters are fitted, the observed centroid proximity between 'attractive' and 'intelligent' faces, for example, could partly reflect proximity of the corresponding prompt embeddings in CLIP's semantic space rather than a visual property of the generated images. The manuscript itself lists CLIP-bias disentanglement only as future work (Section 4, item 1), but this is not a peripheral issue: it concerns the validity of the paper's first main result. I recommend re-running the analysis with an independent image embedding or, better, adding human attractiveness and trait ratings on a sample of the generated images.
  2. [Section 2, prompt design; Section 4, 'we do not define or measure attractiveness'] The paper explicitly states that it 'does not define or measure attractiveness,' yet both main findings treat prompt-derived labels such as 'attractive,' 'unattractive,' 'intelligent,' and 'unintelligent' as ground truth for the perceived facial attribute. No human ratings, face-attribute classifiers, or independent image-level checks confirm that positive-trait faces are actually perceived as positive or that 'unattractive' faces are perceived as unattractive. Without such validation, the measured CLIP distances and classifier error patterns could reflect prompt semantics or other visual correlates rather than the constructs named in the paper.
  3. [Section 3, gender classification results; Section 4, confounding visual cues] The second headline claim attributes higher gender-classifier error rates to 'less-attractive' faces, but the paper itself observes that negative-trait female faces appear older, have neutral or downward expressions, and often lack makeup (Section 4, citing Doh et al. [10]). These are visual confounds that could independently drive classifier errors, as the paper notes when citing Muthukumar et al. [27]. The current experimental design cannot separate 'attractiveness' from age, expression, and makeup, so the statement that classifiers exhibit higher error rates on 'less-attractive' faces goes beyond what the data support without an additional control or covariate analysis.
  4. [Section 2, statistical test; Section 3, significance markers in Figure 2] The description of the statistical test is under-specified. The paper says 'A two-sided t-test was conducted to assess the statistical significance of the centroid distance computed,' but it does not state what the units of comparison are, what the null hypothesis is, or how the 36 cells in Figure 2 are adjusted for multiple comparisons. Since each group has a single centroid, it is unclear whether the test compares per-image distances to the two reference centroids, per-image projections, or something else. Please specify the test procedure precisely, including sample sizes and correction for multiple testing.
minor comments (5)
  1. [Abstract and Section 1] The abstract contains a grammatical error: 'These result raise' should be 'These results raise.' The same phrasing appears in the introduction.
  2. [Section 2, last paragraph] The sentence 'we do not define or measure attractiveness, but focus on analyzing how T2I models associate attractiveness, or it's lack thereof' contains a typo: 'it's' should be 'its.'
  3. [Section 3, DeepFace paragraph] The phrase 'the faces of generated with the negative attributes' is missing a word; it should read 'the faces generated with the negative attributes.'
  4. [References] References [20] and [21] are duplicate entries for the same FairFace paper. Please merge them and renumber.
  5. [Figures 2 and 3] The heatmaps would benefit from an explicit color scale or value labels, since the text relies on visual comparison of small numeric differences. Also, Figure 3 reports means and standard deviations in the caption, but the figure itself does not show error bars; please clarify whether the heatmap cells are means and how variability is displayed.

Circularity Check

0 steps flagged · score 1.0 of 10

No circular derivation: the core measurements are empirical and not fixed by construction; the main caveats are validity limitations, not circularity.

full rationale

The paper's derivation chain contains no fitted parameter that is later renamed as a prediction. The text-to-image result is obtained by generating separate image sets from distinct prompts ('attractive' vs. 'intelligent', etc.), computing CLIP centroids, and measuring Euclidean distances; whether positive-trait centroids fall closer to the 'attractive' centroid is a contingent property of Stable Diffusion 2.1's outputs and is not imposed by the prompts or by any equation. The gender-classification result is measured with three external classifiers (InsightFace, DeepFace, FairFace) and is therefore also empirically independent of the generation step. Self-citations appear (Gulati et al. [14] for trait operationalization; Doh et al. [10] for prior gender-classification findings), but neither is load-bearing: the trait set is additionally grounded in independent social-psychology literature (Dion et al. 1972; Eagly et al. 1991; Todorov & Duchaine 2008), and [10] is cited only as motivation. The explicit limitation in Section 2 that the study 'does not define or measure attractiveness' and the future-work item (1) in Section 4 about disentangling 'biases that could potentially originate from the CLIP embeddings' are construct-validity and measurement-confound caveats, not circularity: the observed CLIP-space association could in principle have failed, and no reduction of the conclusion to the inputs is exhibited. Hence the circularity score is low; a separate rigor critique about prompt-to-attribute validity would target assumptions, not circularity.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

The ledger is small because the paper is an empirical measurement rather than a derivation. The load-bearing assumptions are about measurement validity: prompt words map to traits, CLIP distances map to trait similarity, and prompt gender is ground truth. None of these are independently validated, and the paper acknowledges the CLIP issue as future work.

assumptions (5)
  • domain assumption Prompt words such as 'attractive', 'intelligent', 'trustworthy' produce images that embody those traits as understood by humans.
    The entire similarity analysis in Section 2 treats prompt-labeled image groups as operationalizations of psychological traits; the paper never validates this with human raters.
  • domain assumption Euclidean distance between CLIP embedding centroids captures perceived trait association, with smaller distance meaning stronger association.
    Section 2 defines sim = 1/d and interprets distances as similarity; the authors list disentangling from CLIP embedding bias as future work in Section 4.
  • domain assumption Prompt-specified gender is correct ground truth for measuring gender classification accuracy.
    Section 2 and 3 compute accuracy against the gender word in the generation prompt; synthetic faces may not match the prompt, so 'misclassification' is relative to text labels.
  • domain assumption The attractiveness halo effect literature justifies selecting happiness, sociability, trustworthiness, and intelligence as the traits to test.
    The trait selection in Section 1 relies on prior social psychology findings; the paper does not independently justify why these four traits are the relevant ones.
  • standard math Two-sided t-tests on centroid distances are statistically valid given the sampling procedure.
    The paper reports p < 0.05 but does not specify the unit of analysis, per-image distances versus centroid distances, so the test's validity is unverifiable.

how reviews work

0 comments
Cite this review

Pith. "Pith review of When Algorithms Play Favorites: Lookism in the Generation and Perception of Faces." pith.science (2026). https://pith.science/paper/O5GEG2PY

@misc{pith2026250611025,
  author       = {Pith},
  title        = {Pith review of: When Algorithms Play Favorites: Lookism in the Generation and Perception of Faces},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/O5GEG2PY}},
  note         = {Machine review of arXiv:2506.11025}
}
read the original abstract

This paper examines how synthetically generated faces and machine learning-based gender classification algorithms are affected by algorithmic lookism, the preferential treatment based on appearance. In experiments with 13,200 synthetically generated faces, we find that: (1) text-to-image (T2I) systems tend to associate facial attractiveness to unrelated positive traits like intelligence and trustworthiness; and (2) gender classification models exhibit higher error rates on "less-attractive" faces, especially among non-White women. These result raise fairness concerns regarding digital identity systems.

Figures

Figures reproduced from arXiv: 2506.11025 by the authors.

Figure 1
Figure 1. Examples of the generated faces with Stable Diffusion 2.1 with positive (+) and negative (-) variations for three traits (A [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Heatmaps of centroid distances between attractiveness (A) and other social traits (happiness (H), sociability (S), [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Heatmaps of gender classification accuracy (Mean [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Happy Young Women, Grumpy Old Men? Emotion-Driven Demographic Biases in Synthetic Face Generation

    cs.CY 2026-01 conditional novelty 6.0 of 10

    Emotion words in text-to-image prompts act as demographic selectors: negative emotions shift outputs toward White, middle-aged, male-coded faces, and young Black women are nearly absent across all models.

  2. Filters of Identity: AR Beauty and the Algorithmic Politics of the Digital Body

    cs.HC 2025-06 conditional novelty 4.0 of 10

    AR beauty filters function as technologies of algorithmic governance that enforce racialized, gendered, and ableist beauty standards while concealing their own influence.

Reference graph

Works this paper leans on

41 extracted references · 23 canonical work pages · cited by 2 Pith papers

  1. [10]

    My Kind of Woman

    Miriam Doh et al.2024. “My Kind of Woman": Analysing Gender Stereotypes in AI through The Averageness Theory and EU Law.arXiv preprint arXiv:2407.17474(2024)

    arXiv 2024

  2. [27]

    Varshney

    Vidya Muthukumar, Tejaswini Pedapati, Nalini Ratha, Prasanna Sattigeri, Chai-Wah Wu, Brian Kingsbury, Abhishek Kumar, Samuel Thomas, Aleksandra Mojsilovic, and Kush R. Varshney. 2018. Understanding Unequal Gender Classification Accuracy from Face Images. arXiv:1812.00099 [cs.CV] https://arxiv.org/abs/1812.00099

    arXiv 2018

  3. [1]

    American Psychological Association. n.d.. Ethnicity - APA Dictionary of Psychology. https://dictionary.apa.org/ethnicity Accessed: March 10, 2025. Proceedings of EWAF’25. June 30 – July 02, 2025. Eindhoven, NL. 6•Doh et al

    work page 2025

  4. [2]

    American Psychological Association. n.d.. Race - APA Dictionary of Psychology. https://dictionary.apa.org/race Accessed: March 10, 2025

    work page 2025

  5. [3]

    Mohamed Benkedadra, Dany Rimez, Tiffanie Godelaine, Natarajan Chidambaram, Hamed Razavi Khosroshahi, Horacio Tellez, Matei Mancas, Benoit Macq, and Sidi Ahmed Mahmoudi. 2024. CIA: Controllable Image Augmentation Framework Based on Stable Diffusion. In2024 IEEE 7th International Conference on Multimedia Information Processing and Retrieval (MIPR). IEEE, 600–606

    work page 2024

  6. [4]

    Jasper David Brüns and Martin Meißner. 2024. Do you create your content yourself? Using generative artificial intelligence for social media content creation diminishes perceived brand authenticity.Journal of Retailing and Consumer Services79 (2024), 103790

    work page 2024

  7. [5]

    Joy Buolamwini and Timnit Gebru. 2018. Gender shades: Intersectional accuracy disparities in commercial gender classification. In Conference on fairness, accountability and transparency. PMLR, 77–91

    2018

  8. [6]

    2011.Bodies that matter: On the discursive limits of sex

    Judith Butler. 2011.Bodies that matter: On the discursive limits of sex. routledge

    work page 2011

Show all 41 references
  1. [7]

    Tianwei Chen, Yusuke Hirota, Mayu Otani, Noa Garcia, and Yuta Nakashima. 2024. Would Deep Generative Models Amplify Bias in Future Models?. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 10833–10843

    2024

  2. [8]

    1987.Technologies of Gender: Essays on Theory, Film, and Fiction

    Teresa De Lauretis. 1987.Technologies of Gender: Essays on Theory, Film, and Fiction. Indiana University Press

    1987

  3. [9]

    Karen Dion, Ellen Berscheid, and Elaine Walster. 1972. What is beautiful is good.Journal of Personality and Social Psychology24, 3 (1972), 285–290. https://doi.org/10.1037/h0033731

    1972 doi

  4. [11]

    Eagly, Richard D

    Alice H. Eagly, Richard D. Ashmore, Mona G. Makhijani, and Laura C. Longo. 1991. What is beautiful is good, but...: A meta- analytic review of research on the physical attractiveness stereotype.Psychological Bulletin110, 1 (July 1991), 109–128. https: //doi.org/10.1037/0033-29...

    1991 doi

  5. [12]

    Mast, and Janek S

    Jessika Golle, Fred W. Mast, and Janek S. Lobmaier. 2013. Something to smile about: The interrelationship between attractiveness and emotional expression.Cognition and Emotion28, 2 (July 2013), 298–310. https://doi.org/10.1080/02699931.2013.817383

    2013

  6. [13]

    Aditya Gulati, Bruno Lepri, and Nuria Oliver. 2024. Lookism: The overlooked bias in computer vision.arXiv preprint arXiv:2408.11448 (2024)

    2024 arXiv

  7. [14]

    Aditya Gulati, Marina Martínez-Garcia, Daniel Fernández, Miguel Angel Lozano, Bruno Lepri, and Nuria Oliver. 2024. What is beautiful is still good: the attractiveness halo effect in the era of beauty filters. https://doi.org/10.1098/rsos.240882

    2024 doi

  8. [15]

    Asystematicstudyofbiasamplification

    MelissaHall,LaurensvanderMaaten,LauraGustafson,MaxwellJones,andAaronAdcock.2022. Asystematicstudyofbiasamplification. arXiv preprint arXiv:2201.11706(2022)

    2022 arXiv

  9. [16]

    Phillip Howard, Kathleen C Fraser, Anahita Bhiwandiwalla, and Svetlana Kiritchenko. 2024. Uncovering Bias in Large Vision-Language Models at Scale with Counterfactuals.arXiv preprint arXiv:2405.20152(2024)

    2024 arXiv

  10. [17]

    Julio C. S. Jacques Junior, Cagri Ozcinar, Marina Marjanovic, Xavier Baro, Gholamreza Anbarjafari, and Sergio Escalera. 2019. On the effect of age perception biases for real age regression. In2019 14th IEEE International Conference on Automatic Face & Gesture Recognition (FG 2...

    2019

  11. [18]

    Satoshi Kanazawa and Jody L Kovar. 2004. Why beautiful people are more intelligent.Intelligence32, 3 (2004), 227–243. https: //doi.org/10.1016/j.intell.2004.03.003

    2004 doi

  12. [19]

    Anastasia Karagianni and Miriam Doh. 2024. A feminist legal analysis of non-consensual sexualized deepfakes: contextualizing its impact as AI-generated image-based violence under EU law.Porn Studies0, 0 (2024), 1–18. https://doi.org/10.1080/23268743.2024.2408277 arXiv:https://...

    2024

  13. [20]

    FairFace:Face AttributeDatasetforBalancedRace, Gender,and Age forBias Measurement and Mitigation

    KimmoKarkkainenand Jungseock Joo.2021. FairFace:Face AttributeDatasetforBalancedRace, Gender,and Age forBias Measurement and Mitigation. InProceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV). 1548–1558

    2021

  14. [21]

    FairFace:Face AttributeDatasetforBalancedRace, Gender,and Age forBias Measurement and Mitigation

    KimmoKarkkainenand Jungseock Joo.2021. FairFace:Face AttributeDatasetforBalancedRace, Gender,and Age forBias Measurement and Mitigation. InProceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision. 1548–1558

    2021

  15. [22]

    https://doi.org/10.1145/3442188.3445920

    ZaidKhanandYunFu.2021.OneLabel,OneBillionFaces:UsageandConsistencyofRacialCategoriesinComputerVision.InProceedings ofthe2021ACMConferenceonFairness,Accountability,andTransparency(FAccT’21).ACM. https://doi.org/10.1145/3442188.3445920

    2021

  16. [23]

    Abhishek Kumar, Sarfaroz Yunusov, and Ali Emami. 2024. Subtle Biases Need Subtler Measures: Dual Metrics for Evaluating Represen- tative and Affinity Bias in Large Language Models.arXiv preprint arXiv:2405.14555(2024). Proceedings of EWAF’25. June 30 – July 02, 2025. Eindhoven...

    2024 arXiv

  17. [24]

    Alexandra Sasha Luccioni, Christopher Akiki, Margaret Mitchell, and Yacine Jernite. 2023. Stable bias: Analyzing societal representations in diffusion models.arXiv preprint arXiv:2303.11408(2023)

    2023 arXiv

  18. [25]

    Mathes and Arnold Kahn

    Eugene W. Mathes and Arnold Kahn. 1975. Physical Attractiveness, Happiness, Neuroticism, and Self-Esteem.The Journal of Psychology 90, 1 (May 1975), 27–30. https://doi.org/10.1080/00223980.1975.9923921

    1975

  19. [26]

    Arthur G. Miller. 1970. Role of physical attractiveness in impression formation.Psychonomic Science19, 4 (Oct. 1970), 241–243. https://doi.org/10.3758/bf03328797

    1970 doi

  20. [28]

    Ranjita Naik and Besmira Nushi. 2023. Social biases through the text-to-image generation lens. InProceedings of the 2023 AAAI/ACM Conference on AI, Ethics, and Society. 786–808

    2023

  21. [29]

    LearningTransferableVisualModelsFromNaturalLanguageSupervision

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin,JackClark,GretchenKrueger,andIlyaSutskever.2021. LearningTransferableVisualModelsFromNaturalLanguageSupervision. arXiv:2103.00020 [cs.CV] http...

    2021 arXiv

  22. [30]

    Xingyu Ren, Alexandros Lattas, Baris Gecer, Jiankang Deng, Chao Ma, and Xiaokang Yang. 2023. Facial Geometric Detail Recovery via Implicit Representation. In2023 IEEE 17th International Conference on Automatic Face and Gesture Recognition (FG)

    2023

  23. [31]

    Jonas Ricker, Dennis Assenmacher, Thorsten Holz, Asja Fischer, and Erwin Quiring. 2024. AI-generated faces in the real world: a large-scale case study of twitter profile images. InProceedings of the 27th International Symposium on Research in Attacks, Intrusions and Defenses. 513–530

    2024

  24. [32]

    Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Björn Ommer. 2022. High-Resolution Image Synthesis with Latent Diffusion Models. arXiv:2112.10752 [cs.CV] https://arxiv.org/abs/2112.10752

    2022 arXiv

  25. [33]

    2020), 237802312096717

    CarstenSchwemmer,CarlyKnight,EmilyD.Bello-Pardo,StanOklobdzija,MartijnSchoonvelde,andJeffreyW.Lockhart.2020.Diagnosing Gender Bias in Image Recognition Systems.Socius: Sociological Research for a Dynamic World6 (Jan. 2020), 237802312096717. https://doi.org/10.1177/2378023120967171

    2020 doi

  26. [34]

    Sefik Ilkin Serengil and Alper Ozpinar. 2021. HyperExtended LightFace: A Facial Attribute Analysis Framework. In2021 International Conference on Engineering and Emerging Technologies (ICEET). IEEE, 1–4. https://doi.org/10.1109/ICEET53442.2021.9659697

    2021

  27. [35]

    2016.Perceptions of intelligence and the attractiveness halo

    Sean N Talamas. 2016.Perceptions of intelligence and the attractiveness halo. Ph.D. Dissertation. University of St Andrews

    2016

  28. [36]

    Alexander Todorov and Bradley Duchaine. 2008. Reading trustworthiness in faces without recognizing faces.Cognitive Neuropsychology 25, 3 (May 2008), 395–410. https://doi.org/10.1080/02643290802044996

    2008 doi

  29. [37]

    Amos Tversky and Daniel Kahneman. 1974. Judgment under Uncertainty: Heuristics and Biases.Science185, 4157 (Sept. 1974). https://doi.org/10.1126/science.185.4157.1124

    1974

  30. [38]

    Tianlu Wang, Jieyu Zhao, Mark Yatskar, Kai-Wei Chang, and Vicente Ordonez. 2019. Balanced Datasets Are Not Enough: Estimating and Mitigating Gender Bias in Deep Image Representations. InProceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)

    2019

  31. [39]

    Zeyu Wang, Klint Qinami, Ioannis Christos Karakozis, Kyle Genova, Prem Nair, Kenji Hata, and Olga Russakovsky. 2020. Towards Fairness in Visual Recognition: Effective Strategies for Bias Mitigation. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recogn...

    2020

  32. [40]

    Yankun Wu, Yuta Nakashima, and Noa Garcia. 2023. Stable Diffusion Exposed: Gender Bias from Prompt to Image.arXiv preprint arXiv:2312.03027(2023)

    2023 arXiv

  33. [41]

    Seyma Yucer, Samet Akcay, Noura Al-Moubayed, and Toby P. Breckon. 2020. Exploring Racial Bias Within Face Recognition via Per-Subject Adversarially-Enabled Data Augmentation. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops. ...

    2020

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.