Pith. sign in

REVIEW 3 major objections 7 minor 33 references

LLMs as mirrors of societal moral standards: reflection of cultural divergence and agreement across ethical topics

T0 review · 3 major / 7 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read Six open LLMs probed here mostly fail to mirror cross-cultural moral disagreement, instead projecting a homogenized, Western-leaning moral outlook.

desk verdict Plausible negative result, but the WVS ground-truth preprocessing error makes the central claim unreliable until it is fixed. read the letter →

arxiv 2412.00962 v1 pith:RFB4OE7O submitted 2024-12-01 cs.AI cs.CLcs.SC

Mijntje Meijer , Hadi Mohammadi , Ayoub Bagheri This is my paper · ORCID
classification cs.AIcs.CLcs.SC
keywords largelanguagemodelscross-culturalmoralitymoralvaluesWorldSurveyculturalbiasW.E.I.R.D.prompt-basedprobingvaluepluralism
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 paper asks whether large language models mirror how moral judgments actually differ and agree across countries. Using the World Values Survey Wave 7 and the Pew 2013 Global Attitudes survey as ground truth, the authors probe six open transformer models with templated moral statements and compare the model-generated scores with the survey scores in three ways: cross-topic variance, country clustering, and direct comparative judgments. All three comparisons show weak to near-chance alignment, and the models systematically compress cultural disagreement—they assign higher average moral acceptability and lower variance than the surveys, sometimes listing the most controversial topics (sex before marriage, homosexuality) as among the most agreed upon. The paper concludes that the tested models propagate a homogenized, autonomy-endorsing view consistent with W.E.I.R.D. (Western, Educated, Industrialized, Rich, Democratic) societies, with no convincing advantage for multilingual models or larger parameter counts. A sympathetic reader would care because LLM outputs increasingly mediate global services and public discourse; if the claim is right, these systems systematically under-represent moral pluralism.

What carries the argument

The load-bearing object is the log-probability moral score. For every country-topic pair, the model is prompted to complete 'In {country} {topic} is {moral_judgment}' and 'People in {country} believe {topic} is {moral_judgment}' with five contrasting token pairs—always justifiable/never justifiable, right/wrong, morally good/morally bad, ethically right/ethically wrong, ethical/unethical. The score for a pair is the log probability of the moral token minus the log probability of the immoral token, averaged across the five pairs and both prompt styles, giving one model-generated moral score per country-topic pair that is meant to be comparable to the survey mean. This score feeds all three analyses: Pearson correlation of topic variances, K-means country clustering evaluated by adjusted rand index and adjusted mutual information, and direct comparative prompts that ask whether two countries' judgments are similar or dissimilar. The entire argument depends on these scores being a valid proxy for cultural moral stance.

What would settle it

Run the identical probe suite on a model pretrained predominantly on non-Western, non-English text (e.g., a monolingual Arabic or Hindi model) and check whether the variance correlations, cluster alignments, and probe accuracies against WVS/PEW data exceed the near-zero and chance-level values reported here; if they do, the homogenization is a training-data artifact rather than an inherent LLM limitation. A second test is to re-run the probes with country names replaced by survey-derived cues (e.g., 'in a country where most people say divorce is never justifiable') and see whether the models' moral scores then track the survey variances; if they do, it would indicate the models possess the relevant moral knowledge but the country-prompt format does not retrieve it.

Watch

Extended reading notes

Core claim

The central claim of the paper, stated in the abstract and conclusion, is that the language models tested show 'overall variable and low performance in reflecting cross-cultural differences and similarities in moral values.' Concretely, the correlation between model-generated and survey-based cross-country moral-score variances is weak and mostly negative for the WVS (e.g., r = -0.195 for GPT-2 Medium, r = -0.200 for Qwen) and only moderately positive for the PEW data for GPT-2 Large and BLOOM, without reaching statistical significance. Country clusterings derived from model scores align poorly with survey-derived clusterings, with adjusted rand indices near zero or slightly positive (best Combined Alignment Score 0.215 for Qwen on all WVS topics). Direct comparative probing yields accuracies around 0.5, at or below chance, with some models simply predicting the same class throughout. The paper further claims that the models show a homogenized view: they rate most moral topics as more acceptable and less variable across countries than the surveys do, and they 'generally seem to reflect a rather liberal view, in line with the autonomy-endorsing values found in W.E.I.R.D. societies.' The authors conclude that neither multilinguality nor model size within the tested families convincingly improves this cultural calibration.

Load-bearing premise

The validity of the model-generated moral scores rests on the assumption that asking an English-prompted LLM 'In {country} {topic} is {moral_judgment}' yields a measure of that country's moral stance, rather than a reflection of the model's English-centric training and stereotypes.

Editorial extensions

If this is right

  • If the finding holds, LLMs deployed in global settings will systematically understate how much countries disagree on moral issues, flattening cultural diversity in automated outputs such as search, recommendation, and decision-support systems.
  • The near-chance performance on direct comparative probes implies that even when explicitly asked whether two countries' moral judgments are similar or different, these models do not reproduce survey-observed cultural distances.
  • The models' failure to identify sex before marriage and homosexuality as the most controversial topics suggests particular blind spots on issues where cultural values diverge sharply.
  • The absence of a convincing multilingual or size advantage in these experiments undercuts the common assumption that simply scaling or diversifying pretraining data fixes cultural bias.
  • The results argue for evaluation benchmarks built on ground-truth cross-cultural surveys, and for careful auditing of LLMs before they are used to represent 'what people believe' in a global context.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Beyond the paper's claims: the country-prompt design may itself elicit stereotyped associations rather than moral norms, so a natural extension is to vary the prompt language—presenting the same items in the country's dominant language or with demographic context (age, education, urban/rural)—to test whether model-country scores move closer to survey scores when the measurement frame is less Engli
  • Beyond the paper's claims: the systematic variance compression (higher mean, lower variance) resembles a default to the most frequent training patterns; a directly testable version of this hypothesis is to measure output perplexity or lexical diversity for low-resource countries—if the model's completions for those countries are more generic or higher-perplexity, the compression would track traini
  • Beyond the paper's claims: because the study compares country means, it cannot speak to within-country divides; probing with demographic modifiers (gender, age, education) could reveal whether models reproduce the demographic gradients found in WVS data, providing a more sensitive test of cultural understanding than country means alone.
  • Beyond the paper's claims: the results do not rule out that alignment techniques (instruction tuning, RLHF, or constitution-style training) could improve cultural calibration; a useful next experiment is to fine-tune an open model on survey-style moral judgments with explicit country context and measure whether the probing scores improve.
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

3 major / 7 minor

Summary. The paper asks whether current LLMs reflect cross-cultural moral divergence and agreement, using three methods: comparing variance in moral scores, comparing country clusterings, and probing with direct comparative prompts. The ground truth comes from the WVS Wave 7 and PEW 2013 surveys. Across GPT-2 variants, OPT, Qwen, and BLOOM-family models, the paper reports weak or negative correlations with survey variances, low cluster alignment, and near-chance performance in direct probing, concluding that the models reflect a homogenized, rather liberal, W.E.I.R.D.-leaning view of moral values. The paper is transparent about limitations and does not fit model parameters to the survey data, so the evaluation is external rather than circular.

Significance. If the finding is reliable, it is practically important: it would indicate that current open-weight LLMs do not yet encode empirically observed cross-cultural moral variation, with implications for fairness and global deployment. The paper also makes a useful methodological contribution by combining three complementary probing strategies and two independent survey datasets. Its credibility, however, depends on the integrity of the survey ground truth and on the exact identity of the models tested. The most serious issue is the WVS preprocessing, which in its current form biases the ground-truth scores and therefore undermines the central negative claim. Because that error is fixable and the manuscript is otherwise transparent, the work merits major revision rather than rejection.

major comments (3)
  1. [§4.1] The preprocessing step that replaces WVS response codes -1, -2, -4, and -5 ('Don't know', 'No answer', 'Not asked in survey', 'Missing') with 0 is invalid for the 1–10 justifiability scale. 0 lies below the valid range, so country means are pulled downward; for 'Not asked in survey' the replacement fabricates an extreme 'never justifiable' score for entire country–topic pairs, which inflates cross-country variance and distorts the controversial/agreed topic rankings in Tables 3–4 and 7–8 and the ground-truth clusterings used in Methods 1–3. The statement that replacement with 0 'ensures that non-responses do not influence the computed averages' (Section 4.1) is therefore false. The central negative finding is not reliable until the survey scores are recomputed with these codes excluded or otherwise properly handled.
  2. [§4.2 and all result tables] The methods section states that OPT-350M and BLOOMZ-560M were used (Sections 4.2.1–4.2.2), but every results table lists 'OPT-125' and 'BLOOM' (e.g., Tables 1–2, 11–20). This discrepancy makes it impossible to know which models were actually evaluated and obscures the model-size and multilinguality comparisons that the discussion relies on. Please identify the exact Hugging Face model identifiers and use matching names throughout.
  3. [§4.3] The model-generated country moral scores assume that English prompts of the form 'In {country} {topic} is {moral_judgment}' yield a measure of that country's moral stance. Given that all tested models are predominantly English-trained, the country name may trigger stereotypes or Western-normative associations rather than empirically observed cultural norms. The paper does not validate this scoring method (e.g., against local-language probing or an auxiliary task). Because all three evaluation methods benchmark model scores against survey scores, this is a load-bearing validity assumption. Please discuss this limitation explicitly and, ideally, provide a concrete sensitivity check.
minor comments (7)
  1. [§3.2] The statement that the PEW 2013 survey has '100 participants from each of the 39 countries' appears inconsistent with the typical Pew Global Attitudes sampling of about 1,000 respondents per country; please verify and correct if needed.
  2. [§3.2] The claim that 'The survey questions were given in English' should be clarified, since the Pew Global Attitudes survey is normally administered in local languages.
  3. [§3.1 and §4.1] Section 3.1 describes the data as WVS Wave 7, while Section 4.1 refers to 'version 5 of the World Values Survey (WVS) data'; please clarify whether these refer to the same dataset release.
  4. [§5.3 and Table 20] The text reports a p-value of 0.014 for BLOOM on the PEW data, but Table 20 lists 0.032; also the surrounding text says 'WVS scores' when the table is for the PEW dataset.
  5. [Appendix, Table 37] Table 37 is headed 'Top 5 most agreed on PEW topics according to GPT-2 Large' but lists only three topics.
  6. [§5.1] The text says 'the same two topics ... sex before marriage and homosexuality' are most controversial in both datasets, but the PEW item is 'sex between unmarried adults'; the wording should be aligned with the actual item labels.
  7. [§5.1 and §5.3] The Pearson correlations and chi-square tests are conducted over multiple models and two datasets without multiple-comparison correction; the reported p-values should therefore be interpreted as exploratory, and this should be stated.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: model moral scores come from fixed prompts and token log-probabilities while WVS/PEW survey data serve as external benchmarks; the §4.1 missing-code replacement is a data-quality concern, not a derivation loop.

full rationale

The paper's evaluation chain is not circular. Ground-truth moral scores are computed by averaging survey responses from WVS and PEW (§3.1, §3.2, §4.1), while model-generated moral scores are computed from the log-probability differences of fixed contrasting token pairs under two fixed prompt templates (§4.3). No parameter, prompt, token pair, or scaling constant is fitted to the survey scores. Method 1 compares topic-level variances across countries using Pearson correlation; Method 2 clusters countries with K-means and compares survey and model clusterings with ARI/AMI; Method 3 derives empirical similar/different labels from hierarchical clustering of survey scores and compares them with model judgments. In all three methods the survey data are external targets rather than ingredients of the model score computation. Reusing the survey-selected K for the model clustering is a design decision that does not inject survey values into model outputs or mathematically force alignment. The §4.1 practice of replacing WVS codes -1, -2, -4, -5 with 0 before averaging may bias the ground-truth means and affect the validity of the benchmark, but this is a data-quality and measurement issue that applies equally to the external target; it does not make the model's scores depend on the survey values, nor does it rename a fitted parameter as a prediction. There is no load-bearing self-citation, no imported uniqueness theorem, and no ansatz smuggled in via citation; the central negative finding is an empirically falsifiable comparison against an independent dataset.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The paper introduces no new physical or conceptual entities; its evaluation pipeline rests on survey aggregation assumptions, log-probability scoring assumptions, and the non-neutral replacement of missing survey responses with 0. These are choices, not fitted parameters in a predictive model.

free parameters (3)
  • Non-response replacement value = 0
    In Section 4.1, WVS responses coded -1,-2,-4,-5 are replaced with 0 before averaging; 0 is outside the valid 1-10 response scale and pulls country means downward, distorting subsequent normalization and variance estimates.
  • K for k-means clustering = varies (1-10 chosen by silhouette score)
    In Section 4.4.2, K is selected from [1,10] by maximizing silhouette score on survey scores, then reused for model scores; K affects ARI/AMI alignment values.
  • K for hierarchical clustering in direct probing = selected by elbow method per topic
    In Section 4.4.3, the number of clusters for country grouping per topic is chosen using the elbow method, affecting which two clusters are compared.
assumptions (4)
  • domain assumption Country-level means of survey responses are a valid ground truth for cross-cultural moral values
    Used throughout; acknowledged as a limitation in Section 7 because averaging obscures within-country diversity.
  • domain assumption Log-probability differences between contrasting moral tokens measure the model's moral judgment
    The moral score in Section 4.3 is defined as the averaged log-probability difference; no validation is provided that this correlates with human judgments.
  • ad hoc to paper The five token pairs and two prompt styles provide an unbiased estimate of moral stance
    These specific prompts and token pairs are chosen by the authors without evidence of equivalence across languages and cultures; prompt sensitivity is acknowledged as a limitation in Section 7.
  • ad hoc to paper Replacing non-responses with 0 is neutral
    In Section 4.1, the replacement is claimed not to influence averages, but 0 is outside the valid range and is not neutral; this assumption is false.

how reviews work

0 comments
Cite this review

Pith. "Pith review of LLMs as mirrors of societal moral standards: reflection of cultural divergence and agreement across ethical topics." pith.science (2026). https://pith.science/paper/RFB4OE7O

@misc{pith2026241200962,
  author       = {Pith},
  title        = {Pith review of: LLMs as mirrors of societal moral standards: reflection of cultural divergence and agreement across ethical topics},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/RFB4OE7O}},
  note         = {Machine review of arXiv:2412.00962}
}
read the original abstract

Large language models (LLMs) have become increasingly pivotal in various domains due the recent advancements in their performance capabilities. However, concerns persist regarding biases in LLMs, including gender, racial, and cultural biases derived from their training data. These biases raise critical questions about the ethical deployment and societal impact of LLMs. Acknowledging these concerns, this study investigates whether LLMs accurately reflect cross-cultural variations and similarities in moral perspectives. In assessing whether the chosen LLMs capture patterns of divergence and agreement on moral topics across cultures, three main methods are employed: (1) comparison of model-generated and survey-based moral score variances, (2) cluster alignment analysis to evaluate the correspondence between country clusters derived from model-generated moral scores and those derived from survey data, and (3) probing LLMs with direct comparative prompts. All three methods involve the use of systematic prompts and token pairs designed to assess how well LLMs understand and reflect cultural variations in moral attitudes. The findings of this study indicate overall variable and low performance in reflecting cross-cultural differences and similarities in moral values across the models tested, highlighting the necessity for improving models' accuracy in capturing these nuances effectively. The insights gained from this study aim to inform discussions on the ethical development and deployment of LLMs in global contexts, emphasizing the importance of mitigating biases and promoting fair representation across diverse cultural perspectives.

Figures

Figures reproduced from arXiv: 2412.00962 by the authors.

Figure 1
Figure 1. Distribution of normalized answer values for [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Spread of responses across the moral topics [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Distribution of normalized answer values for [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Spread of responses across the moral topics [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: Comparison between the degrees of cultural diversities and shared tendencies in the empirical moral [PITH_FULL_IMAGE:figures/full_fig_p010_5.png]
Figure 6
Figure 6. Figure 6: Comparison between the degrees of cultural diversities and shared tendencies in the empirical moral [PITH_FULL_IMAGE:figures/full_fig_p010_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

33 extracted references · 8 canonical work pages

  1. [1]

    URL: " 'urlintro :=

    ENTRY address author booktitle chapter edition editor howpublished institution journal key month note number organization pages publisher school series title type volume year eprint doi pubmed url lastchecked label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block STRINGS urlintro eprinturl eprintpr...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

  3. [3]

    Arora, L

    A. Arora, L. Kaffee, and I. Augenstein. 2022. https://doi.org/10.48550/arxiv.2203.13722 Probing pre-trained language models for cross-cultural differences in values . arXiv preprint arXiv:2203.13722

    arXiv doi:10.48550/arxiv.2203.13722 2022

  4. [4]

    Bender, Timnit Gebru, Angelina McMillan-Major, and Shmargaret Shmitchell

    Emily M. Bender, Timnit Gebru, Angelina McMillan-Major, and Shmargaret Shmitchell. 2021. https://doi.org/10.1145/3442188.3445922 On the dangers of stochastic parrots: Can language models be too big? Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency, pages 610--623

    work page arXiv 2021

  5. [5]

    Yochai Benkler, Danny Mosaphir, Sarah Friedman, Alice Smart, and Stefan Schmer-Galunder. 2023. https://doi.org/10.48550/arxiv.2312.10075 Assessing llms for moral value pluralism . arXiv (Cornell University)

    arXiv doi:10.48550/arxiv.2312.10075 2023

  6. [6]

    Zou, Venkatesh Saligrama, and Adam T

    Tolga Bolukbasi, Kai-Wei Chang, James Y. Zou, Venkatesh Saligrama, and Adam T. Kalai. 2016. https://doi.org/10.48550/arxiv.1606.06121 Quantifying and reducing stereotypes in word embeddings . arXiv

    arXiv doi:10.48550/arxiv.1606.06121 2016

  7. [7]

    Joy Buolamwini and Timnit Gebru. 2018. https://dl.acm.org/doi/10.1145/3287560.3287596 Gender shades: Intersectional accuracy disparities in commercial gender classification . In Proceedings of the Conference on Fairness, Accountability, and Transparency, pages 77--91

    arXiv 2018

  8. [8]

    Gaëlle Cachat-Rosset and Alain Klarsfeld. 2023. https://doi.org/10.1080/08839514.2023.2176618 Diversity, equity, and inclusion in artificial intelligence: An evaluation of guidelines . Applied Artificial Intelligence, 37(1)

    arXiv 2023

Show all 33 references
  1. [9]

    Bryson, and Arvind Narayanan

    Aylin Caliskan, Joanna J. Bryson, and Arvind Narayanan. 2017. https://doi.org/10.1126/science.aal4230 Semantics derived automatically from language corpora contain human-like biases . Science, 356(6334):183--186

    2017 doi

  2. [10]

    Xinrun Du, Zhouliang Yu, Songyang Gao, Ding Pan, Yuyang Cheng, Ziyang Ma, Ruibin Yuan, Xingwei Qu, Jiaheng Liu, Tianyu Zheng, Xinchen Luo, Guorui Zhou, Binhang Yuan, Wenhu Chen, Jie Fu, and Ge Zhang. 2024. https://doi.org/10.48550/arxiv.2404.04167 Chinese Tiny LLM: Pretraining...

    2024 doi arXiv

  3. [11]

    Luciano Floridi, Josh Cowls, Monica Beltrametti, Raimondi Chatila, Prabhat Agarwal, Mariarosaria Bianchi, Virginia Bodei, Emiliano Bonfiglio, Massimo Calvaresi, Alberto Chersac, Mario Chiaro, Jessica De Schutter, Nicola Di Caro, Franco Di Rollo, Donato A. Fabrizio, Caterina Ga...

    2018

  4. [12]

    Johnson, and Liane Zhang

    Jesse Graham, Peter Meindl, Eliza Beall, Kate M. Johnson, and Liane Zhang. 2016. https://doi.org/10.1016/j.copsyc.2015.09.007 Cultural differences in moral judgment and behavior, across and within societies . Current Opinion in Psychology, 8:125--130

    2016 doi

  5. [13]

    Haerpfer, Patrick Bernhagen, Ronald F

    Christian W. Haerpfer, Patrick Bernhagen, Ronald F. Inglehart, and Christian Welzel. 2022. http://www.worldvaluessurvey.org/WVSDocumentationWV7.jsp World Values Survey: Round Seven - Country-Pooled Datafile Version . Institute for Comparative Survey Research, Vienna

    2022

  6. [14]

    Jonathan Haidt. 2001. https://doi.org/10.1037/0033-295X.108.4.814 The emotional dog and its rational tail: A social intuitionist approach to moral judgment . Psychological Review, 108(4):814--834

    2001 doi

  7. [15]

    R. L. Johnson, G. Pistilli, N. Menédez-González, L. D. D. Duran, E. Panai, J. Kalpokiene, and D. J. Bertulfo. 2022. https://arxiv.org/abs/2203.07785 The ghost in the machine has an american accent: value conflict in gpt-3 . arXiv.org

    2022 arXiv

  8. [16]

    Kostas Karpouzis. 2024. https://doi.org/10.3390/electronics13081457 Plato’s shadows in the digital cave: Controlling cultural bias in generative ai . Electronics, 13(8):1457

    2024 doi

  9. [17]

    Kharchenko, T

    J. Kharchenko, T. Roosta, A. Chadha, and C. Shah. 2024. http://arxiv.org/abs/2406.14805 How well do llms represent values across cultures? empirical analysis of llm responses based on hofstede cultural dimensions . arXiv preprint. ArXiv:2406.14805

    2024 arXiv

  10. [18]

    Denys Lazarenko and Thomas Bonald. 2021. http://arxiv.org/abs/2103.12641 Pairwise adjusted mutual information

    2021 arXiv

  11. [19]

    Teven Le Scao, Angela Fan, Christopher Akiki, Ellie Pavlick, Suzana Ili\'c, Daniel Hesslow, Roman Castagn\'e, Alexandra Sasha Luccioni, Fran c ois Yvon, Matthias Gall \'e , Jonathan Tow, Alexander M. Rush, Stella Biderman, Albert Webson, Pawan Sasanka Ammanamanchi, Thomas Wang...

    2022 arXiv

  12. [20]

    C. C. Liu, F. Koto, T. Baldwin, and I. Gurevych. 2023. http://arxiv.org/abs/2309.08591 Are multilingual llms culturally-diverse reasoners? an investigation into multicultural proverbs and sayings . arXiv. ArXiv:2309.08591

    2023 arXiv

  13. [21]

    Ninareh Mehrabi, Fred Morstatter, Nripsuta Saxena, Kristina Lerman, and Aram Galstyan. 2021. https://doi.org/10.1145/3447926 A survey on bias and fairness in machine learning . ACM Computing Surveys, 54(6):1--35

    2021 doi

  14. [22]

    Ashish Mishra, Gyanaranjan Nayak, Suparna Bhattacharya, Tarun Kumar, Arpit Shah, and Martin Foltin. 2024. https://doi.org/10.1145/3589335.3651929 Llm-guided counterfactual data generation for fairer ai . In Companion Proceedings of the ACM on Web Conference 2024, WWW '24, page...

    2024

  15. [23]

    Tanya Nazaretsky, Sara Hershkovitz, and Giora Alexandron. 2020. https://eric.ed.gov/?id=ED599209 Kappa learning: A new item-similarity method for clustering educational items from response data

    2020

  16. [24]

    Praveen Nemani, Yu David Joel, Priya Vijay, and Farrah F. Liza. 2024. https://doi.org/10.1016/j.nlp.2023.100047 Gender bias in transformers: A comprehensive review of detection and mitigation strategies . Natural Language Processing Journal, 6:100047

    2024

  17. [25]

    Safiya Umoja Noble. 2018. https://nyupress.org/9781479837243/algorithms-of-oppression/ Algorithms of Oppression: How Search Engines Reinforce Racism . NYU Press, New York, NY

    2018

  18. [26]

    Alec Radford, Jeff Wu, Rewon Child, David Luan, Dario Amodei, and Ilya Sutskever. 2019. https://api.semanticscholar.org/CorpusID:160025533 Language models are unsupervised multitask learners

    2019

  19. [27]

    Atefeh Ramezani and Yuhan Xu. 2023. https://doi.org/10.48550/arxiv.2306.01857 Knowledge of cultural moral norms in large language models . arXiv (Cornell University)

    2023 doi arXiv

  20. [28]

    Danny Matthew Saputra, Daniel Saputra, and Liniyanti D. OSWARI. 2020. https://doi.org/10.2991/aisr.k.200424.051 Effect of distance metrics in determining k-value in k-means clustering using elbow and silhouette method . In Proceedings of the Sriwijaya International Conference ...

    2020 doi

  21. [29]

    big three

    Richard A. Shweder, Nancy C. Much, Manamohan Mahapatra, and Lawrence Park. 1997. https://psycnet.apa.org/record/1997-36447-005 The "big three" of morality (autonomy, community, divinity) and the "big three" explanations of suffering . In A. Brandt and P. Rozin, editors, Morali...

    1997

  22. [30]

    L. Wang, X. Chen, and X. Deng. 2024. https://doi.org/10.1038/s41746-024-01029-4 Prompt engineering in consistency and reliability with the evidence-based guideline for llms . npj Digital Medicine, 7:41

    2024 doi

  23. [31]

    Susan Zhang, Stephen Roller, Naman Goyal, Mikel Artetxe, Moya Chen, Shuohui Chen, Christopher Dewan, Mona Diab, Xian Li, Xi Victoria Lin, et al. 2022. Opt: Open pre-trained transformer language models. arXiv preprint arXiv:2205.01068

    2022 arXiv

  24. [32]

    James Zou and Londa Schiebinger. 2018. https://doi.org/10.1038/d41586-018-05707-8 Ai can be sexist and racist—it’s time to make it fair . Nature, 559(7714):324--326

    2018 doi

  25. [33]

    Didar Zowghi and De Rosa Francesca. 2023. https://doi.org/10.48550/arxiv.2305.12728 Diversity and inclusion in artificial intelligence . arXiv. ArXiv:2305.12728

    2023 doi arXiv

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.