Wisdom from Diversity: Bias Mitigation Through Hybrid Human-LLM Crowds

Axel Abels; Tom Lenaerts

REVIEW 4 major objections 5 minor 2 cited by

Wisdom from Diversity: Bias Mitigation Through Hybrid Human-LLM Crowds

T0 review · 4 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read Hybrid crowds that mix humans and LLMs, combined through locally weighted aggregation, can judge bias-eliciting headlines more accurately than either group alone while reducing demographic bias to near zero.

desk verdict First head-to-head test of hybrid human-LLM crowds for bias mitigation; the qualitative story is solid, but the headline 'significant' claim lacks paired uncertainty. read the letter →

arxiv 2505.12349 v1 pith:6U74O6SB submitted 2025-05-18 cs.CL cs.AIcs.CYcs.HCcs.LG

Axel Abels , Tom Lenaerts This is my paper · ORCID

classification cs.CLcs.AIcs.CYcs.HCcs.LG

keywords biasmitigationhybridhuman-LLMcrowdswisdomofExpertiseTreeslocallyweightedaggregationcounterfactualLLMensemblescollectiveintelligence

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 claims that the most reliable way to reduce demographic bias in language-model judgments is not to add more language models, but to add a few humans. On a set of genuine and altered news headlines, individual LLMs reproduce the same counterfactual biases humans show, yet LLMs are far more correlated with one another than humans are, so averaging LLM responses can lock in shared mistakes. The paper shows that locally weighted aggregation, which gives each responder context-dependent weight, recovers much of the wisdom of the crowd. The strongest result is that hybrid crowds of humans and LLMs, aggregated this way, reach higher accuracy than either group alone while leaving counterfactual bias near zero.

What carries the argument

The load-bearing object is the ExpertiseTree, a decision-tree-like aggregation method whose leaves contain linear combinations of individual predictions rather than constants. Each leaf learns weights for a specific context region—here, headline category (age, ethnicity, gender)—so a responder who is reliable for gender headlines but biased on ethnicity headlines can be weighted up in one region and down in the other. Combined with a hybrid crowd, this lets the aggregate exploit LLM accuracy where LLMs are strong and human diversity where LLMs share the same blind spots. A second load-bearing component is the Q-statistic, which quantifies pairwise agreement and motivates why simple averaging fails when responders are highly correlated.

What would settle it

Run the same ExpertiseTree hybrid aggregation on an independent bias-eliciting headline dataset built around a different cultural context (e.g., caste or regional ethnicity) with a human pool of several hundred; if the hybrid crowd does not beat both human-only and LLM-only ExpertiseTrees while keeping counterfactual bias near zero, the claim is refuted. A cheaper check: measure the average Q-statistic of a newly released set of LLMs; if it has fallen to human levels (~0.4), the diversity argument for hybrids loses its empirical basis.

Watch

Extended reading notes

Core claim

The central discovery is a complementarity: LLMs bring higher individual accuracy and humans bring diversity, and the two together, when combined with context-sensitive weighting, outperform either alone. The paper measures diversity with the Q-statistic and finds LLM pairs far more similar (Q=0.855 ± 0.08) than human pairs (Q=0.387 ± 0.33), so simple averaging of LLM outputs barely improves accuracy and can even strengthen biases, as seen in the gender-altered aggregate. ExpertiseTrees, which learn separate linear combination weights for each headline category, route around biased or inaccurate responders. On the headline task, ExpertiseTree(hybrid+) reaches 0.813 accuracy with counterfactual bias deltas within ±0.04 for all reported categories, versus 0.752 for the LLM-only equivalent and 0.749 for the human-only equivalent. The paper concludes that modest hybrid ensembles can combine the accuracy of LLMs with the diversity of human perspectives to achieve more accurate and less biased collective judgments.

Load-bearing premise

The comparison rests on the headline dataset and the roughly 40 human participants from the earlier study being representative of real-world demographic bias; if those headlines or that participant pool do not generalize to other cultural contexts, the measured hybrid advantage may not hold.

Editorial extensions

If this is right

Simple averaging of multiple LLM responses is not a reliable bias-mitigation strategy; because LLMs are highly correlated with each other, averaging can amplify a bias that only a minority of models show.
Locally weighted aggregation (ExpertiseTrees) improves both accuracy and bias metrics over simple and static weighted averaging for LLM, human, and hybrid crowds.
Hybrid crowds of humans and LLMs matched or beat single-type crowds in accuracy while reducing counterfactual bias to levels no longer significant in the tested categories.
Group-size gains differ by composition: LLM-only groups plateau or decline as models are added, while hybrid groups continue to benefit from added humans because each human adds diversity, not redundancy.
Selecting LLMs by MMLU score (LLM+, hybrid+) improves aggregate performance further, but even with such selection, adding more LLMs without humans eventually degrades performance.

Reading between the lines

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

Because the human pool in the underlying dataset is small (around 40 participants), the quantitative superiority of hybrids is likely tied to that pool's particular mix of views; a much larger or differently selected pool could shift the balance.
The same design likely transfers to other judgment tasks where LLMs share systematic blind spots—content moderation, medical or legal triage—but only if the task has a comparable counterfactual structure that lets locally weighted aggregation learn where humans add diversity.
As LLM families converge on shared training data and alignment procedures, the Q-statistic between models should be monitored; if it rises toward 1, the hybrid advantage will depend even more on recruiting humans with genuinely distinct perspectives.
A testable extension is deliberately engineering LLM diversity through personas or fine-tuning and asking whether that substitutes for, or only augments, human diversity in the aggregate.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

First thing to know: this is the first head-to-head comparison I've seen of human-only, LLM-only, and hybrid crowds on a bias-eliciting task, and the headline result—hybrid crowds with locally weighted aggregation beat both pure crowds—is a real new data point. The secondary finding that averaging LLM outputs amplifies shared biases is also concrete and likely correct.

What the paper does well: it diagnoses why LLM crowds fail to self-correct—average pairwise Q-statistic among LLMs is 0.855 vs 0.387 for humans, so errors are correlated. It then shows that ExpertiseTrees, which learn context-specific weights, recover a good deal of the crowd benefit. The experimental setup is clean, the code is public, and the limitations section is honest about the single dataset and model versions.

Where I'd push back: the central claim of 'significantly enhance' is anchored on Table 1, where ExpertiseTree(hybrid+) scores 0.813 vs 0.752 for LLM+ and 0.749 for human. Those numbers are reported without paired confidence intervals or a significance test. Given the pool is ~40 humans and 18 LLMs, and the ExpertiseTree weights are fit with cross-validation on the same task, a 0.061 gap could plausibly be noise. The near-zero bias deltas for hybrid+ also lack intervals. Figure 2 shows CIs for group size trends, but not for the specific group-8 comparisons in Table 1. So the 'significantly' is not yet established. That's the main soft spot.

Second, the aggregation weights are fitted to the same headline dataset. Cross-validation prevents gross overfitting, but this remains an in-sample demonstration. A careful referee should ask for a held-out task or at least a discussion of when this recipe transfers.

These are fixable, not fatal. The qualitative pattern—LLM diversity is low, hybrid diversity helps, simple averaging amplifies bias—is well supported by the Q-statistic analysis and the group-size curves. I'd trust that direction.

Who's this for? Researchers in collective intelligence, AI fairness, and LLM evaluation. It's a solid empirical paper that deserves a serious referee; I'd send it out and ask for paired uncertainty quantification and an out-of-sample check before accepting. I'd also want the authors to soften 'significantly' until the intervals are there.

Referee Report

4 major / 5 minor

Summary. The paper studies bias mitigation in a headline authenticity task by aggregating responses from human and LLM crowds. It first replicates a human headline experiment with 18 LLMs, showing that LLMs mirror human counterfactual biases, especially for ethnicity-related headlines. It then evaluates simple averaging, weighted averaging, and locally weighted ExpertiseTree aggregation for LLM-only, human-only, and hybrid crowds. The central claim is that hybrid crowds containing both humans and LLMs, aggregated with ExpertiseTrees, achieve the highest accuracy (0.813 for ExpertiseTree(hybrid+) in Table 1) and reduce counterfactual bias to near-zero levels, outperforming both human-only and LLM-only groups. The paper also analyzes crowd diversity via the Q-statistic and reports that simple LLM averaging can amplify biases.

Significance. If the central result holds, the paper offers a practical, empirically grounded recipe for combining human and LLM responses to improve accuracy and reduce social biases in content moderation and similar tasks. It contributes to the collective intelligence literature by providing evidence that humans and LLMs have complementary strengths (diversity vs. accuracy) and that locally weighted aggregation can exploit this complementarity. The study is strengthened by its open-source code, the systematic evaluation of 18 diverse LLMs, and the use of appropriate significance tests for the per-cell counterfactual bias comparisons. However, the main quantitative claim—that hybrid crowds significantly outperform LLM-only and human-only groups—is currently under-supported because the headline accuracies in Table 1 are reported without paired confidence intervals or paired significance tests, and the accuracy threshold for the Likert-to-binary mapping is unspecified.

major comments (4)

[§4.2, Table 1, Figure 2] The central claim that ExpertiseTree(hybrid+) 'significantly enhance[s] performance' rests on point estimates (0.813 vs 0.752 for LLM+ and 0.749 for human) that are reported without paired uncertainty estimates. Group construction introduces multiple sources of sampling variability: random draws of 8 humans from a pool of 40, random LLM subsets for the non-plus groups, and cross-validation folds for weight fitting. Figure 2 shows 95% confidence intervals by group size, but the paper never states whether the specific group-size-8 comparisons have non-overlapping intervals or whether the paired difference excludes zero. I request bootstrap or paired permutation confidence intervals for the accuracy and bias-delta comparisons in Table 1, particularly for ExpertiseTree(hybrid+) versus ExpertiseTree(LLM+) and ExpertiseTree(human).
[§3.2, Accuracy definition] Accuracy is defined as 'the proportion of correctly identified headlines, whether genuine or altered,' but the paper does not specify how a 5-point Likert likelihood response is converted to a binary correct/incorrect decision. A threshold (e.g., p_h >= 0.5) is necessary to compute every accuracy number in Table 1 and Figure 2. Without this threshold, the headline results are not reproducible. Please state the threshold explicitly and provide a sensitivity analysis over plausible thresholds to confirm that the hybrid advantage is not an artifact of the chosen cutoff.
[§3.3 and §S1.2 (ExpertiseTrees)] The ExpertiseTree aggregator is central to the paper's positive results, but the supplementary description lacks algorithmic detail: no pseudocode, no specification of tree depth, minimum leaf size, split criterion, pruning, or regularization, and no description of how cross-validation is nested with the random group sampling. Because the method originates in the authors' prior work, these details may exist elsewhere, but the paper should state the exact configuration used here for independent verification. Additionally, since the method and the human comparison data come from the authors' own prior papers and the weights are fitted on the same headline task via cross-validation, the paper should address the risk that the method's advantage reflects tuning on this specific dataset; a nested model-selection or an evaluation on an independent task would materially strengthen the generality claim.
[Limitations and Conclusion] The paper is commendable for including a Limitations section that acknowledges the single-dataset scope, but the abstract and conclusion make broad claims: 'hybrid crowds containing both significantly enhance performance and further reduce biases across ethnic and gender-related contexts.' Given one dataset and one pool of 40 human participants, these statements should be explicitly scoped to the studied task and participant pool, or supported by at least one additional dataset or an independent human sample. This is not a fundamental flaw because the limitation is acknowledged, but the strength of the wording exceeds what the evidence currently supports.

minor comments (5)

[Abstract] The abstract uses 'significantly enhance performance' without reporting any significance test for the hybrid advantage; this wording should be aligned with the statistical evidence (or revised after adding paired tests).
[§3.2 and Table 1] The paper should report the number of headlines in each category-sentiment demographic cell, since the Mann-Whitney U and Wilcoxon tests have low power with very small cell sizes; this context is essential for interpreting the cell shading.
[Table 1 and §3.1] The terms 'Black' and 'African American' are used interchangeably; please choose one and use it consistently.
[References] The reference to 'Hendrycks et al.' lacks a complete citation (year and venue); please provide the full bibliographic entry.
[Figure 2 caption] Please clarify in the caption whether the shaded areas are 95% confidence intervals for the mean accuracy or for the distribution of group draws, and state how many random group draws were used to compute them.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the hybrid-crowd claim rests on new LLM data, external MMLU-based selection, and cross-validated ExpertiseTree evaluation, not on a self-citation chain or a fitted quantity renamed as a prediction.

full rationale

The paper's central derivation chain is self-contained in the relevant sense. The headline dataset and human responses come from prior work by the same authors, but that is data provenance, not circularity: the new LLM responses are freshly elicited, the MMLU benchmark used for LLM+ and hybrid+ selection is external, and the ExpertiseTree weights are obtained by cross-validation ('For both weighted average methods, we use cross-validation to ensure weights were not trained on headlines they are evaluated on'), so the reported accuracies and bias deltas are not fitted inputs being re-reported as predictions. The counterfactual-bias and framing-effect definitions are standard measurement equations, not definitions of the claimed outcome, and no equation in the paper reduces to another by construction. The self-citations to ExpertiseTrees and the headline experiment are methodological building blocks rather than unverified premises used to force the conclusion. The absence of paired confidence intervals for the Table 1 accuracy gap is a statistical robustness concern, not a circularity. The paper itself acknowledges the single-dataset limitation, which again concerns external validity rather than circular reasoning.

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

The central claim rests on the headline dataset, the human response pool from the authors' prior work, and the ExpertiseTree weights fitted to the task. No new entities are introduced.

free parameters (3)

ExpertiseTree leaf weights w_m^phi(h) = Not reported (learned per category via cross-validation)
The local aggregation weights for each responder/model and headline category are fitted to the headline dataset, and the bias mitigation and accuracy gains depend on them.
ExpertiseTree split structure = Not reported (data-driven)
The decision-tree splits over context (age, gender, ethnicity) are selected by optimizing performance on the same task.
Group size (8) for headline comparisons = 8
Table 1 reports groups of 8; other sizes are explored in Figure 2. This is an experimental design choice, not a theory parameter.

assumptions (4)

domain assumption Headline labels (genuine vs altered) are reliable ground truth for accuracy.
Accuracy is computed as the proportion of correctly identified headlines using these labels.
domain assumption Counterfactual bias and framing effects are valid operationalizations of social bias.
Section 3.2 defines bias via likelihood differences across demographic groups; this assumes such differences indicate bias rather than legitimate base-rate differences.
domain assumption The human responses from Abels et al. 2024 are representative of human bias and diversity.
The human pool of roughly 40 participants is the sole source of human diversity and bias for the hybrid crowd claims.
standard math The Q-statistic is an appropriate measure of ensemble diversity.
Yule's Q is a standard measure of classifier agreement, used to quantify diversity in Section 3.2.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Wisdom from Diversity: Bias Mitigation Through Hybrid Human-LLM Crowds." pith.science (2026). https://pith.science/paper/6U74O6SB

@misc{pith2026250512349,
  author       = {Pith},
  title        = {Pith review of: Wisdom from Diversity: Bias Mitigation Through Hybrid Human-LLM Crowds},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6U74O6SB}},
  note         = {Machine review of arXiv:2505.12349}
}

read the original abstract

Despite their performance, large language models (LLMs) can inadvertently perpetuate biases found in the data they are trained on. By analyzing LLM responses to bias-eliciting headlines, we find that these models often mirror human biases. To address this, we explore crowd-based strategies for mitigating bias through response aggregation. We first demonstrate that simply averaging responses from multiple LLMs, intended to leverage the "wisdom of the crowd", can exacerbate existing biases due to the limited diversity within LLM crowds. In contrast, we show that locally weighted aggregation methods more effectively leverage the wisdom of the LLM crowd, achieving both bias mitigation and improved accuracy. Finally, recognizing the complementary strengths of LLMs (accuracy) and humans (diversity), we demonstrate that hybrid crowds containing both significantly enhance performance and further reduce biases across ethnic and gender-related contexts.

Figures

Figures reproduced from arXiv: 2505.12349 by the authors.

**Figure 2.** A demonstrates that locally weighted averages de [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗

**Figure 1.** A. Model accuracy across prompt variations, with each dot representing the accuracy of a model for a specific prompt ending. B. t-SNE visualization of responder diversity. C. Q-statistics (see Equation 2) matrix. Each cell gives the Q-statistic between responder pairs. A B [PITH_FULL_IMAGE:figures/full_fig_p007_1.png] view at source ↗

Figures from the paper (1 more)

**Figure 2.** Figure 2: Accuracy for different group sizes and aggregator [PITH_FULL_IMAGE:figures/full_fig_p007_2.png]

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

Proof2Hybrid: Automatic Mathematical Benchmark Synthesis for Proof-Centric Problems
cs.CL 2025-08 conditional novelty 7.0 of 10

A fully automated pipeline produces proof-centric math benchmarks, demonstrated on algebraic geometry with 456 items, where leading LLMs score near 60 percent.
Floor, Ceiling, and the Fusion Gap: How Much of Crowd Reading Attention Can Machines Predict?
cs.IR 2026-08 conditional novelty 6.0 of 10

Machines recover up to 53% of the crowd-highlight prediction headroom, and fusing five frontier models reaches about 60%, confirmed in a pre-registered replication.

Reference graph

Works this paper leans on

20 extracted references · 8 canonical work pages · cited by 2 Pith papers

[3]

Qwen technical report

Jinze Bai, Shuai Bai, Y unfei Chu, Zeyu Cui, Kai Dang, Xiaodong Deng, Y ang Fan, Wenbin Ge, Y u Han, Fei Huang, et al. Qwen technical report. arXiv preprint arXiv:2309.16609,

arXiv
[14]

The woman worked as a babysitter: On biases in language generation

Emily Sheng, Kai-Wei Chang, Prem Natarajan, and Nanyun Peng. The woman worked as a babysitter: On biases in language generation. In Proceedings of the 2019 Confer- ence on Empirical Methods in Natural Language Process- ing and the 9th International Joint Conference on Natu- ral Language Processing (EMNLP-IJCNLP), pages 3407– 3412,

work page 2019
[15]

Towards controllable biases in language generation

Emily Sheng, Kai-Wei Chang, Prem Natarajan, and Nanyun Peng. Towards controllable biases in language generation. In Findings of the Association for Computational Linguis- tics: EMNLP 2020 , pages 3239–3254,

work page 2020
[18]

Gender bias in coreference resolu- tion: Evaluation and debiasing methods

Jieyu Zhao, Tianlu Wang, Mark Y atskar, Vicente Ordonez, and Kai-Wei Chang. Gender bias in coreference resolu- tion: Evaluation and debiasing methods. In Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Lan- guage T echnologies, volume 2,

work page 2018
[20]

As an AI model, I

75, 1}. Responses which failed to answer the query, such as those containing system-level disclaimers (“As an AI model, I. . . ”) were re-queried for consistency. S1.2 ExpertiseTrees ExpertiseTrees [Abels et al., 2023] are an advanced aggre- gation method designed to partition the context space (e.g. , headline categories) and ﬁt localized aggregations fo...

work page 2023
[1947]

Efﬁcient estimation of word representation s in vector space

Tomas Mikolov. Efﬁcient estimation of word representation s in vector space. arXiv preprint arXiv:1301.3781 , 3781,

arXiv
[1990]

Gemini: a family of highly capable multimodal models

Team Gemini, Rohan Anil, Sebastian Borgeaud, Jean- Baptiste Alayrac, Jiahui Y u, Radu Soricut, Johan Schalk- wyk, Andrew M Dai, Anja Hauth, Katie Millican, et al. Gemini: a family of highly capable multimodal models. arXiv preprint arXiv:2312.11805 ,

arXiv
[1996]

Lan- guage models are few-shot learners

Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, et al. Lan- guage models are few-shot learners. Advances in neural information processing systems, 33:1877–1901,

1901

Show all 20 references

[2000]

Wisdom of the silicon crowd: Llm en- semble prediction capabilities match human crowd accu- racy

Philipp Schoenegger, Indre Tuminauskaite, Peter S Park, an d Philip E Tetlock. Wisdom of the silicon crowd: Llm en- semble prediction capabilities match human crowd accu- racy. arXiv preprint arXiv:2402.19379 ,

arXiv
[2004]

Multimodal datasets: misogyny, pornography, and malignant stereotypes

Abeba Birhane, Vinay Uday Prabhu, and Emmanuel Kahem- bwe. Multimodal datasets: misogyny, pornography, and malignant stereotypes. arXiv preprint arXiv:2110.01963 ,

arXiv
[2005]

Evaluating and mitigating dis- crimination in language model decisions

Alex Tamkin, Amanda Askell, Liane Lovitt, Esin Durmus, Nicholas Joseph, Shauna Kravec, Karina Nguyen, Jared Kaplan, and Deep Ganguli. Evaluating and mitigating dis- crimination in language model decisions. arXiv preprint arXiv:2312.03689,

arXiv
[2013]

Mistral chat: Advanced conversational ai, 2023

MistralAI. Mistral chat: Advanced conversational ai, 2023 . Available at https://mistral.ai. Devah Pager. Marked: Race, crime, and ﬁnding work in an era of mass incarceration . University of Chicago Press,

2023
[2016]

On the opportunities and risks of foundation models

Rishi Bommasani, Drew A Hudson, Ehsan Adeli, Russ Alt- man, Simran Arora, Sydney von Arx, Michael S Bernstein, Jeannette Bohg, Antoine Bosselut, Emma Brunskill, et al. On the opportunities and risks of foundation models. arXiv preprint arXiv:2108.07258,

arXiv
[2017]

The llama 3 herd of models

Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Ab- hishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Amy Y ang, Angela Fan, et al. The llama 3 herd of models. arXiv preprint arXiv:2407.21783 ,

arXiv
[2018]

Note that several models are proprietary, and their sizes are therefore estimated based on their cost, their re- sponse times, as well as their performance

Supplementary Information for Wisdom from Diversity: Bias Mitigation Through Hybrid Huma n-LLM Crowds S1 Supplementary Information Table S2 presents an overview of the LLMs we considered in this work, including the version, estimates of the number of parameters, their MMLU [He...

2024
[2019]

On the dangers of stochastic parrots: Can language models be too big? In Proceedings of the 2021 ACM conference on fairness, accountability, and transparency, pages 610–623,

Emily M Bender, Timnit Gebru, Angelina McMillan-Major, and Shmargaret Shmitchell. On the dangers of stochastic parrots: Can language models be too big? In Proceedings of the 2021 ACM conference on fairness, accountability, and transparency, pages 610–623,

2021
[2020]

Men also like shopping: Reducing gender bias ampliﬁcation using corpus-level constraints

Jieyu Zhao, Tianlu Wang, Mark Y atskar, Vicente Ordonez, and Kai-Wei Chang. Men also like shopping: Reducing gender bias ampliﬁcation using corpus-level constraints. In Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing . Association for Com...

2017
[2021]

Deepseek-v3 technical report

Aixin Liu, Bei Feng, Bing Xue, Bingxuan Wang, Bochao Wu, Chengda Lu, Chenggang Zhao, Chengqi Deng, Chenyu Zhang, Chong Ruan, et al. Deepseek-v3 technical report. arXiv preprint arXiv:2412.19437 ,

arXiv
[2023]

Mitigating biases in collective decision-makin g: Enhancing performance in the face of fake news

Axel Abels, Elias Fernandez Domingos, Ann Now´ e, and Tom Lenaerts. Mitigating biases in collective decision-makin g: Enhancing performance in the face of fake news. arXiv preprint arXiv:2403.08829,

arXiv
[2024]

Gpt-4 technical report

Josh Achiam, Steven Adler, Sandhini Agarwal, Lama Ah- mad, Ilge Akkaya, Florencia Leoni Aleman, Diogo Almeida, Janko Altenschmidt, Sam Altman, Shyamal Anadkat, et al. Gpt-4 technical report. arXiv preprint arXiv:2303.08774,

arXiv

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.

[1] [3]

Qwen technical report

Jinze Bai, Shuai Bai, Y unfei Chu, Zeyu Cui, Kai Dang, Xiaodong Deng, Y ang Fan, Wenbin Ge, Y u Han, Fei Huang, et al. Qwen technical report. arXiv preprint arXiv:2309.16609,

arXiv

[2] [14]

The woman worked as a babysitter: On biases in language generation

Emily Sheng, Kai-Wei Chang, Prem Natarajan, and Nanyun Peng. The woman worked as a babysitter: On biases in language generation. In Proceedings of the 2019 Confer- ence on Empirical Methods in Natural Language Process- ing and the 9th International Joint Conference on Natu- ral Language Processing (EMNLP-IJCNLP), pages 3407– 3412,

work page 2019

[3] [15]

Towards controllable biases in language generation

Emily Sheng, Kai-Wei Chang, Prem Natarajan, and Nanyun Peng. Towards controllable biases in language generation. In Findings of the Association for Computational Linguis- tics: EMNLP 2020 , pages 3239–3254,

work page 2020

[4] [18]

Gender bias in coreference resolu- tion: Evaluation and debiasing methods

Jieyu Zhao, Tianlu Wang, Mark Y atskar, Vicente Ordonez, and Kai-Wei Chang. Gender bias in coreference resolu- tion: Evaluation and debiasing methods. In Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Lan- guage T echnologies, volume 2,

work page 2018

[5] [20]

As an AI model, I

75, 1}. Responses which failed to answer the query, such as those containing system-level disclaimers (“As an AI model, I. . . ”) were re-queried for consistency. S1.2 ExpertiseTrees ExpertiseTrees [Abels et al., 2023] are an advanced aggre- gation method designed to partition the context space (e.g. , headline categories) and ﬁt localized aggregations fo...

work page 2023

[6] [1947]

Efﬁcient estimation of word representation s in vector space

Tomas Mikolov. Efﬁcient estimation of word representation s in vector space. arXiv preprint arXiv:1301.3781 , 3781,

arXiv

[7] [1990]

Gemini: a family of highly capable multimodal models

Team Gemini, Rohan Anil, Sebastian Borgeaud, Jean- Baptiste Alayrac, Jiahui Y u, Radu Soricut, Johan Schalk- wyk, Andrew M Dai, Anja Hauth, Katie Millican, et al. Gemini: a family of highly capable multimodal models. arXiv preprint arXiv:2312.11805 ,

arXiv

[8] [1996]

Lan- guage models are few-shot learners

Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, et al. Lan- guage models are few-shot learners. Advances in neural information processing systems, 33:1877–1901,

1901