Pith. sign in

REVIEW 4 major objections 7 minor 63 references

Examining Alignment of Large Language Models through Representative Heuristics: The Case of Political Stereotypes

T0 review · 4 major / 7 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read Large language models systematically exaggerate US party positions beyond human survey respondents, in a pattern consistent with representativeness heuristics.

desk verdict The descriptive finding (LLMs exaggerate partisan positions) is solid and worth knowing, but the representativeness-heuristic mechanism as tested is circular and should be reframed or fixed. read the letter →

arxiv 2501.14294 v3 pith:HZ7HOHWO submitted 2025-01-24 cs.CL cs.AI

Sullam Jeoung , Yubin Ge , Haohan Wang , Jana Diesner This is my paper · ORCID
classification cs.CLcs.AI
keywords representativenessheuristicspoliticalstereotypinglargelanguagemodelsalignmentkerneloftruthlikelihoodratiopromptmitigationsurveyresponsebias
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 tries to show that when large language models are asked to place the Democratic and Republican parties on political attitude scales, their answers land further from the empirical center than human survey respondents' guesses do: Republican positions are inflated, Democratic positions deflated, and the gap between the two is stretched. The authors borrow the cognitive-science idea of representativeness heuristics, where people exaggerate attributes that are diagnostically distinctive of a group, and test whether LLM responses obey the same formal pattern. They find a kernel of truth: model predictions correlate with real partisan differences, then amplify them. If the claim holds, LLM outputs about US politics are not merely left- or right-leaning but systematically stereotyped, and the same measurement lens could expose equivalent distortions for other groups.

What carries the argument

The load-bearing formal object is the representativeness ratio R[a] = p(a|X+)/p(a|X-), the likelihood that a scale position occurs for Republicans versus Democrats. The paper defines the most representative attribute, the exemplar, as the argmax of this ratio, defines kappa as the ratio between the model's conditional probability at the exemplar and its representativeness, and uses two linear equations to express the predicted mean as the empirical mean plus an epsilon-weighted excess of tail representativeness. Positive epsilon is taken as evidence that the model overweights representative attributes when forming predictions.

What would settle it

Simulate synthetic LLM responses by resampling the empirical partisan distributions with added noise but no representativeness weighting, then fit the paper's two linear equations to the simulated means; if positive epsilon appears as often as in the real model outputs, the representativeness mechanism is not identified by this test. Alternatively, hold out half of each model's responses to compute the representativeness ratio and the other half to compute the predicted mean; if epsilon collapses, the effect is an artifact of same-distribution regression.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is a quantitative profile of LLM political stereotyping. Across seven models and two survey instruments, asking a model 'where would you place the Democratic/Republican Party' yields mean responses that consistently overshoot the empirical means of self-identified partisans: Republican-associated predictions are too high, Democratic-associated predictions are too low, and predicted party differences exceed both empirical differences and human predictions. The deviations fit the kernel-of-truth equation (positive gamma), meaning predictions track real group differences while stretching them, and the representativeness equations yield positive epsilon for most model-topic combinations, which the authors read as evidence that likelihood-ratio representativeness, not just random noise, shapes the exaggeration. Prompt strategies modeled on human debiasing, such as awareness, feedback, and reasoning instructions, reduce the kappa measure of stereotyping but do not eliminate it.

Load-bearing premise

The mechanism claim rests on treating deviations of the predicted mean from the empirical mean as evidence of representativeness when the representativeness ratio is computed from the same model-output distribution that defines the predicted mean, and no null model shows that positive epsilon cannot arise mechanically from that circularity.

Editorial extensions

If this is right

  • LLM-generated survey-style outputs about US parties are systematically more polarized than the underlying empirical positions, so downstream systems that consume such outputs inherit exaggerated party differences.
  • Because the exaggeration exceeds human predictive bias, alignment evaluations that compare models to human average guesses will understate the distortion.
  • The kernel-of-truth finding implies the distortion is not arbitrary: models reproduce real partisan correlations and then amplify them, so mitigation cannot simply erase the signal.
  • Prompt-based self-correction can lower the measured stereotyping proxy but not remove it, suggesting the bias is partly baked into the model rather than purely a surface prompt effect.
  • The same likelihood-ratio lens can be applied to any paired target and reference group to test whether the exaggeration pattern is general.

Reading between the lines

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

  • An untested implication is that the same protocol would show analogous exaggeration for non-political groups such as gender, nationality, or ethnicity, since the likelihood-ratio formalism is group-agnostic and the training-data mechanisms are not politics-specific.
  • The authors' own comparison of an instruction-tuned model with its base version hints that alignment training may increase rather than decrease this exaggeration; a direct test would compare base and instruction-tuned checkpoints of the same model on the full battery.
  • A practical extension is to use the epsilon coefficient as a monitor for political stereotyping during model deployment, recalibrating prompts or filtering outputs when epsilon drifts upward.
  • The downstream misinformation experiment is only suggestive; a causal test would manipulate the representativeness of the provided party context and measure changes in truth judgments.
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 / 7 minor

Summary. The paper studies whether LLMs' predictions of US political party positions (Republicans, Democrats) on ANES and MFQ survey items deviate systematically from empirical human self-placements and from human predictions of party positions. The authors formalize stereotypes via the representativeness heuristic: an attribute is representative if its likelihood ratio P(a|X+)/P(a|X-) is high. They fit a kernel-of-truth parameter γ (Eq. 4) and a representativeness parameter ε (Eqs. 5–6), report that LLM predictions are more polarized than both empirical means and human predictions, that ε is often positive, and that prompt-based mitigation strategies (AWARENESS, REASONING, FEEDBACK) reduce κ (Eq. 3). The paper concludes that LLMs are susceptible to representativeness heuristics, which facilitates political stereotyping.

Significance. If the results hold, the paper would provide a theoretically grounded, cross-model characterization of a specific form of political bias in LLMs—systematic exaggeration of partisan positions beyond human exaggeration—and would connect it to a well-known cognitive heuristic. The descriptive part is useful and relatively robust: across six LLMs and two datasets, predicted means for Republicans are consistently higher than empirical means and human predictions, and for Democrats consistently lower. The paper ships code and uses public survey data, which supports reproducibility. However, the mechanism claim (that the exaggeration is due to representativeness heuristics) rests on fitted parameters ε, γ, and κ that, as implemented, are mathematically coupled to the very distributions they are supposed to explain. The paper provides no null model, permutation test, or confidence intervals for these parameters. Thus the paper's headline contribution—'LLMs overemphasize representativeness more than humans'—is not yet established, even though the raw exaggeration pattern is credible.

major comments (4)
  1. [§3, Eqs. (5)–(6), Appendix C] The representativeness test is circular in the current implementation. The exemplar a* in Eq. (2) and the right-tail set A(N) are computed from the model's own response distribution p^B_{a,X+}/p^B_{a,X-}, and the predicted mean E^B(a|X+) is also computed from that same p^B. Eq. (5) therefore does not relate an independent representativeness measure to a predicted mean; it solves for ε as (E^B(a|X+) − E(a|X+))/(P^L_{A(N)} − 1), which is an algebraic identity given the definitions. Any model whose distribution is shifted toward high-scale values relative to the empirical distribution will produce positive ε whenever the right tail has above-average representativeness in the model's own distribution. A model with no heuristic—say, one that anchors on endpoints, repeats a default extreme, or has miscalibrated variance—would yield the same sign pattern. The same concern applies to the κ measure in Eq. (3), which also uses p^B, and to the Human_Pred comparison, since the same construction is applied to human predictions. The manuscript reports only means and standard deviations across topics (Table 14) and no significance test, null model, or permutation baseline. The conclusion that 'LLMs tend to overemphasize representativeness more than humans' (Abstract) is therefore not supported by the evidence as presented.
  2. [§6, Table 1 and Table 2] The fitted parameters γ and ε are reported without uncertainty quantification. Table 1 reports γ values with standard deviations in parentheses, but no significance tests, confidence intervals, or model comparison against a null model (e.g., γ=0). Table 2 reports ε with standard deviations, but again no test of whether ε differs from zero, let alone whether LLM ε exceeds Human_Pred ε. Given that several entries are within one standard deviation of zero (e.g., Llama2-70b ANES R ε = −0.84±4.96, Gpt-4 ANES R ε = −0.08±2.60), the claim of systematic positive ε is not statistically supported. The paper should provide confidence intervals or bootstrap/permutation tests for ε and γ, and should test whether the LLM ε distribution is stochastically larger than the Human_Pred ε distribution.
  3. [§6, Table 3] The mitigation analysis uses κ (Eq. 3) as the outcome measure, but κ is computed from the same model distribution p^B that the prompting interventions are expected to change. If ε is mechanically positive for shifted distributions, then a prompt that flattens the distribution or moves it toward the empirical mean will mechanically reduce κ, regardless of whether the intervention actually targets a representativeness heuristic. The claim that the mitigation strategies are 'effective in reducing the influence of representativeness' (Abstract) therefore conflates a distributional shift with a reduction in heuristic use. A proper test would compare the intervention's effect on ε or κ against a null model where the intervention merely rescales or shifts the distribution without changing its tail representativeness.
  4. [§6, Table 1; §10 Limitations] The kernel-of-truth analysis (Eq. 4) has a sign-direction issue that is acknowledged only in a footnote. The paper states that Eq. 4 holds 'if and only if the group has a higher average position than the other group' (footnote 3), but the empirical means in ANES do not always satisfy this ordering (e.g., Women's Rights: Republican empirical mean 2.83 vs Democratic 2.56; the prediction for Republicans is higher, but for Democrats the mean difference is small). For topics where the ordering is reversed or where the two means are close, γ can become large or negative for mechanical reasons, as seen in Table 13 (e.g., Llama2-70b γ = 4.18 for Women's Rights R, but also γ = −1.14 for Legal Rights R). The paper should filter or report which topics satisfy the ordering condition, or use a more general formulation that does not require the ordering assumption. Without this, the kernel-of-truth γ is not a clean measure of 'inflation'.
minor comments (7)
  1. [Abstract and §1] The phrase 'the extent and conditions under which LLMs deviate from empirical positions are insufficiently examined' is repeated almost verbatim in the Abstract and Introduction; consider varying the wording.
  2. [§3, Eq. (2) and Appendix A] The exemplar definition and the Laplace smoothing are described in the main text and appendix, but the value of N in A(N) is set to 2 in the main text and then the appendix says 'We set N = 2 for our analysis' earlier; please make the N selection justification explicit (e.g., robustness checks for N=1,3).
  3. [§5.1 and Table 4] The ANES sample sizes in Table 4 are very large (often >10,000 per party per topic), but the LLM responses are based on 20 repetitions per prompt. The comparison of a near-deterministic LLM (some models have zero variance in Table 12) with a large human sample should be described as a comparison of point estimates, not of distributions, and the paper should avoid implying that a small number of LLM samples can be compared to human sampling variability.
  4. [Appendix E, Table 6] The temperature sensitivity analysis reports CV and mean differences, but does not report the standard deviation of the Diff_D and Diff_R across temperatures; the claim that 'the deviation from the empirical mean remains relatively consistent' would be stronger with a measure of dispersion around the mean difference.
  5. [Appendix G, Table 10] The misinformation detection experiment is labeled as preliminary, but the accuracy values for some cells (e.g., 0.739 for Llama2-70b Democrats with party) appear to be single-point estimates without error bars. Please add confidence intervals or note the small sample sizes for Democrat-labeled statements.
  6. [Throughout] The paper alternates between 'Republican' and 'Democrat' labels and 'X+'/'X−' notation inconsistently in tables and figures. For instance, Table 2 uses R and D, but Eqs. (5)–(6) use ε_{X+} and ε_{X−}; please define the correspondence clearly in each table caption.
  7. [References] The reference list is missing several recent works that would help contextualize the contribution, such as studies on political bias in LLMs beyond the cited ones (e.g., 'Whose opinions do language models reflect?' is cited as Santurkar et al. 2023, but the citation style is inconsistent). Please conduct a final reference check.

Circularity Check

1 steps flagged · score 6.0 of 10

The representativeness-heuristic test in Eqs. 5–6 is definitionally coupled to the predicted mean: the right-tail representativeness P is computed from the same model response distribution that defines E^B, so positive ε is an algebraic consequence, not independent evidence of a heuristic. The raw polarization finding is not circular.

  1. self definitional [Section 3, Eq. 5 (and Eq. 6), with definitions of p^B, the exemplar (Eq. 2), and P^X+_{A(N)}; Appendix C documents the 20-response sampling]
    "a ∗ ∈ arg max a pB a,X+ / pB a,X− (2) ... We denote P X+ A(N ) = (Σ A(N ) pa,X+) / (Σ A(N ) pa,X−) as the average representativeness of the right tail. ... EB(a|X +) = E(a|X +) + ϵX + · (P X+ A(N ) − 1) (5) ... Equations 5 and 6 measure the degree to which the representativeness accounted for forming the predicted mean. If ϵX + > 0 and ϵX − > 0, we assume the Predicted Mean exhibits representative heuristics, positively weighting the representativeness."

    P^X+_{A(N)} is built from the model's own response distribution p^B (via Eq. 2 and the right-tail definition), while E^B(a|X+) is the mean of that same p^B. Eq. 5 therefore does not relate an independent representativeness measure to a predicted mean; it defines ε = (E^B − E)/(P^L − 1), a ratio of two functionals of one distribution. Any model whose right tail has an above-average likelihood ratio and whose mean lies toward that tail yields ε > 0 mechanically, even without any representativeness heuristic (e.g., simple endpoint anchoring or variance miscalibration would produce the same sign pattern).

full rationale

The descriptive finding is not circular: LLM party-placement means are compared against external ANES/MFQ empirical means and against human predictions, and Figure 4 shows an amplification above the diagonal that does not depend on the heuristic formalism. The kernel-of-truth γ (Eq. 4) and κ (Eq. 3) are descriptive ratios, with κ anchored to empirical probabilities, so those components are largely independent of the circularity. However, the mechanism claim in the abstract and Section 6—that LLMs 'tend to overemphasize representativeness more than humans'—rests on Eqs. 5–6, where the right-tail representativeness P is computed from the same model response distribution p^B whose mean appears on the left side. As a result, positive ε values can arise from generic distributional properties, and no null model is provided to rule out the mechanical explanation. The self-citations (Jeoung et al. 2023a,b) are background and do not carry the load-bearing argument. Overall, the raw exaggeration result is self-contained, but the representativeness-heuristic evidence is partially circular by construction, warranting a score of 6.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The paper's quantitative framework rests on two imported cognitive-science models (Eq 4 and Eqs 5-6) and on treating 20 repeated LLM responses as a stable probability distribution. The central exaggeration result is a direct empirical comparison, but the representativeness mechanism is measured by fitting coefficients to the same data they explain. No new physical or theoretical entities are introduced.

free parameters (4)
  • gamma (kernel-of-truth coefficient) = Table 1: e.g., Gpt-3.5 ANES 1.66; Llama3-8b MFQ -0.33
    Fit per model and dataset to Eq 4; it transforms empirical party means into predicted means. The paper presents positive gamma as evidence for kernel-of-truth, but it is fitted to the same data it describes.
  • epsilon_X+ and epsilon_X- (representativeness weights) = Table 2: e.g., Gemini ANES Democrats 4.00; Llama2-70b ANES Republicans -0.84
    Fit via Eqs 5-6 to quantify how much of the predicted mean deviation is attributed to the right-tail representativeness ratio computed from the model's own outputs.
  • N (right-tail size) = N=2
    Chosen by hand in Section 3 ('We set N = 2 for our analysis') with no sensitivity analysis; affects P_A(N) and hence all epsilon estimates.
  • Laplace smoothing coefficient alpha = alpha=1 (add-one)
    Applied in Appendix A to avoid zero probabilities when estimating R from 20 model responses; small-sample smoothing changes likelihood ratios materially.
assumptions (5)
  • domain assumption Belief formation follows Eq 4 from Bordalo et al.: predicted mean = (1+gamma) times empirical mean minus gamma times contrast mean.
    Borrowed from cognitive science and imposed as the model of kernel-of-truth; not derived for LLMs. Used in Section 3.
  • domain assumption Representativeness of an attribute is the likelihood ratio p(a|X+)/p(a|X-), and the right tail (top N ratios) drives stereotype exaggeration.
    Standard cognitive-science definition, but the right-tail top-N operationalization is an additional modeling choice made in Section 3.
  • domain assumption LLM response frequencies from 20 repeated generations approximate the model's subjective probability distribution p^B.
    The paper estimates all p^B from 20 samples (Appendix C); this assumes sampling noise and smoothing do not distort R, kappa, or epsilon.
  • domain assumption Self-identified party affiliation in ANES and MFQ is a valid proxy for ideological position, and empirical survey means are the correct ground truth for LLM predictions.
    Acknowledged in the Limitations section, but it is load-bearing for every deviation claim.
  • domain assumption Party conditioning prompts elicit the model's belief about the party, not a stylized text-completion artifact or refusal behavior.
    The entire prediction task depends on this; Appendix G shows response-ratio issues in some models, indicating this assumption is not always clean.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Examining Alignment of Large Language Models through Representative Heuristics: The Case of Political Stereotypes." pith.science (2026). https://pith.science/paper/HZ7HOHWO

@misc{pith2026250114294,
  author       = {Pith},
  title        = {Pith review of: Examining Alignment of Large Language Models through Representative Heuristics: The Case of Political Stereotypes},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/HZ7HOHWO}},
  note         = {Machine review of arXiv:2501.14294}
}
read the original abstract

Examining the alignment of large language models (LLMs) has become increasingly important, e.g., when LLMs fail to operate as intended. This study examines the alignment of LLMs with human values for the domain of politics. Prior research has shown that LLM-generated outputs can include political leanings and mimic the stances of political parties on various issues. However, the extent and conditions under which LLMs deviate from empirical positions are insufficiently examined. To address this gap, we analyze the factors that contribute to LLMs' deviations from empirical positions on political issues, aiming to quantify these deviations and identify the conditions that cause them. Drawing on findings from cognitive science about representativeness heuristics, i.e., situations where humans lean on representative attributes of a target group in a way that leads to exaggerated beliefs, we scrutinize LLM responses through this heuristics' lens. We conduct experiments to determine how LLMs inflate predictions about political parties, which results in stereotyping. We find that while LLMs can mimic certain political parties' positions, they often exaggerate these positions more than human survey respondents do. Also, LLMs tend to overemphasize representativeness more than humans. This study highlights the susceptibility of LLMs to representativeness heuristics, suggesting a potential vulnerability of LLMs that facilitates political stereotyping. We also test prompt-based mitigation strategies, finding that strategies that can mitigate representative heuristics in humans are also effective in reducing the influence of representativeness on LLM-generated responses.

Figures

Figures reproduced from arXiv: 2501.14294 by the authors.

Figure 1
Figure 1. An example from the ANES survey. Responses from self-identified Democrats and Republicans human partici￾pants Empirical Question are denoted as Em￾pirical and the answers generated by LLMs to Prediction Questions as Prediction. Second, we study whether the responses generated by LLMs exhibit representative heuristics by evaluating the extent to which LLMs exhibit such shortcuts in their outputs. This investigation s… view at source ↗
Figure 4
Figure 4. The x-axis corresponds to the Empirical Mean Difference (E(a|X +) − E(a|X −)), and the y-axis corresponds to the Predicted Mean Difference (E B(a|X +) − E B(a|X −)) of each question. The black line indicates y = x. Our analysis of responses to MFQ reveals systematic pat￾terns of distortion across political affiliations. For Demo￾cratic positions, LLMs generally demonstrate a downward bias, producing predictions belo… view at source ↗
Figure 2
Figure 2. Analysis ofANES Response Distributions. Response distributions are presented using mean scales with associated ranges. Data points represent mean values, while error bars indicate the range of observed responses. The Empirical Mean represents average scores from self-identified Democrats and Republicans (corresponding to Empirical Question in [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figures from the paper (3 more)
Figure 3
Figure 3. Figure 3: Results of MFQ Responses. The figure presents the deviation between LLM-generated MFQ responses and Empirical Mean values across political affiliations. Republican-associated predictions show predominantly positive differences across most LLMs, indicating systematic ov…
Figure 5
Figure 5. Figure 5: The ANES responses, categorized by topics. Empirical represents the average scale from self￾identified Democrats and Republicans (on Empirical Question in [PITH_FULL_IMAGE:figures/full_fig_p018_5.png]
Figure 6
Figure 6. Figure 6: The MFQ analysis results, categorized by topics. Empirical represents the average scale from self-identified Democrats and Republicans from Empirical Questions. Full numerical mean and std details are available in Appendix 12. ANES Women’s Rights Urban Unrest Legal Rig…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

63 extracted references · 30 canonical work pages

  1. [1]

    Gpt-4 technical report

    Josh Achiam, Steven Adler, Sandhini Agarwal, Lama Ahmad, Ilge Akkaya, Florencia Leoni Aleman, Diogo Almeida, Janko Altenschmidt, Sam Altman, Shyamal Anadkat, et al. Gpt-4 technical report. arXiv preprint arXiv:2303.08774, 2023

    arXiv 2023

  2. [2]

    Llama 3 model card

    AI@Meta. Llama 3 model card. 2024. URL https://github.com/meta-llama/llama3/blob/main/MODEL_CARD.md

    2024

  3. [3]

    Out of one, many: Using language models to simulate human samples

    Lisa P Argyle, Ethan C Busby, Nancy Fulda, Joshua R Gubler, Christopher Rytting, and David Wingate. Out of one, many: Using language models to simulate human samples. Political Analysis, 31 0 (3): 0 337--351, 2023

    2023

  4. [4]

    A general language assistant as a laboratory for alignment

    Amanda Askell, Yuntao Bai, Anna Chen, Dawn Drain, Deep Ganguli, Tom Henighan, Andy Jones, Nicholas Joseph, Ben Mann, Nova DasSarma, et al. A general language assistant as a laboratory for alignment. arXiv preprint arXiv:2112.00861, 2021

    arXiv 2021

  5. [5]

    Errors in probabilistic reasoning and judgment biases

    Daniel J Benjamin. Errors in probabilistic reasoning and judgment biases. Handbook of Behavioral Economics: Applications and Foundations 1, 2: 0 69--186, 2019

    work page 2019

  6. [6]

    Using cognitive psychology to understand gpt-3

    Marcel Binz and Eric Schulz. Using cognitive psychology to understand gpt-3. Proceedings of the National Academy of Sciences, 120 0 (6): 0 e2218523120, 2023

    work page 2023

  7. [7]

    Stereotyping norwegian salmon: An inventory of pitfalls in fairness benchmark datasets

    Su Lin Blodgett, Gilsinia Lopez, Alexandra Olteanu, Robert Sim, and Hanna Wallach. Stereotyping norwegian salmon: An inventory of pitfalls in fairness benchmark datasets. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Pape...

    2021

  8. [8]

    Man is to computer programmer as woman is to homemaker? debiasing word embeddings

    Tolga Bolukbasi, Kai-Wei Chang, James Y Zou, Venkatesh Saligrama, and Adam T Kalai. Man is to computer programmer as woman is to homemaker? debiasing word embeddings. Advances in neural information processing systems, 29, 2016

    work page 2016

Show all 63 references
  1. [9]

    Stereotypes

    Pedro Bordalo, Katherine Coffman, Nicola Gennaioli, and Andrei Shleifer. Stereotypes. The Quarterly Journal of Economics, 131 0 (4): 0 1753--1794, 2016

    2016

  2. [10]

    Race gaps in sat scores highlight inequality and hinder upward mobility, 2017

    Brookings. Race gaps in sat scores highlight inequality and hinder upward mobility, 2017. URL https://www.brookings.edu/articles/race-gaps-in-sat-scores-highlight-inequality-and-hinder-upward-mobility

    2017

  3. [11]

    Language 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. Language models are few-shot learners. Advances in neural information processing systems, 33: 0 1877--1901, 2020

    1901

  4. [12]

    Theory-grounded measurement of us social stereotypes in english language models

    Yang Cao, Anna Sotnikova, Hal Daum \'e III, Rachel Rudinger, and Linda Zou. Theory-grounded measurement of us social stereotypes in english language models. In Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Hu...

    2022

  5. [13]

    Can large language models understand content and propagation for misinformation detection: An empirical study

    Mengyang Chen, Lingwei Wei, Han Cao, Wei Zhou, and Songlin Hu. Can large language models understand content and propagation for misinformation detection: An empirical study. arXiv preprint arXiv:2311.12699, 2023

    2023 arXiv

  6. [14]

    From pretraining data to language models to downstream tasks: Tracking the trails of political biases leading to unfair NLP models

    Shangbin Feng, Chan Young Park, Yuhan Liu, and Yulia Tsvetkov. From pretraining data to language models to downstream tasks: Tracking the trails of political biases leading to unfair NLP models. In Proceedings of the 61st Annual Meeting of the Association for Computational Lin...

    2023 doi

  7. [15]

    Misinfo reaction frames: Reasoning about readers ' reactions to news headlines

    Saadia Gabriel, Skyler Hallinan, Maarten Sap, Pemi Nguyen, Franziska Roesner, Eunsol Choi, and Yejin Choi. Misinfo reaction frames: Reasoning about readers ' reactions to news headlines. In Smaranda Muresan, Preslav Nakov, and Aline Villavicencio (eds.), Proceedings of the 60t...

    2022 doi

  8. [16]

    Gallup. U.s. 1\ URL https://news.gallup.com/poll/151310/u.s.-republican-not-conservative.aspx

  9. [17]

    The capacity for moral self-correction in large language models

    Deep Ganguli, Amanda Askell, Nicholas Schiefer, Thomas Liao, Kamil \.e Luko s i \=u t \.e , Anna Chen, Anna Goldie, Azalia Mirhoseini, Catherine Olsson, Danny Hernandez, et al. The capacity for moral self-correction in large language models. arXiv preprint arXiv:2302.07459, 2023

    2023 arXiv

  10. [18]

    Moral foundations theory: The pragmatic validity of moral pluralism

    Jesse Graham, Jonathan Haidt, Sena Koleva, Matt Motyl, Ravi Iyer, Sean P Wojcik, and Peter H Ditto. Moral foundations theory: The pragmatic validity of moral pluralism. In Advances in experimental social psychology, volume 47, pp.\ 55--130. Elsevier, 2013

    2013

  11. [19]

    Bayes rule as a descriptive model: The representativeness heuristic

    David M Grether. Bayes rule as a descriptive model: The representativeness heuristic. The Quarterly journal of economics, 95 0 (3): 0 537--557, 1980

    1980

  12. [20]

    The political ideology of conversational ai: Converging evidence on chatgpt’s pro-environmental, left-libertarian orientation

    Jochen Hartmann, Jasper Schwenzow, and Maximilian Witte. The political ideology of conversational ai: Converging evidence on chatgpt’s pro-environmental, left-libertarian orientation. Left-Libertarian Orientation (January 1, 2023), 2023

    2023

  13. [21]

    Stereotypes

    James L Hilton and William Von Hippel. Stereotypes. Annual review of psychology, 47 0 (1): 0 237--271, 1996

    1996

  14. [22]

    Examining the causal impact of first names on language models: The case of social commonsense reasoning

    Sullam Jeoung, Jana Diesner, and Halil Kilicoglu. Examining the causal impact of first names on language models: The case of social commonsense reasoning. In Proceedings of the 3rd Workshop on Trustworthy Natural Language Processing (TrustNLP 2023), pp.\ 61--72, 2023 a

    2023

  15. [23]

    S tereo M ap: Quantifying the awareness of human-like stereotypes in large language models

    Sullam Jeoung, Yubin Ge, and Jana Diesner. S tereo M ap: Quantifying the awareness of human-like stereotypes in large language models. In Houda Bouamor, Juan Pino, and Kalika Bali (eds.), Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, p...

    2023 doi

  16. [24]

    Communitylm: Probing partisan worldviews from language models

    Hang Jiang, Doug Beeferman, Brandon Roy, and Deb Roy. Communitylm: Probing partisan worldviews from language models. In Proceedings of the 29th International Conference on Computational Linguistics, pp.\ 6818--6826, 2022

    2022

  17. [25]

    Definition and assessment of accuracy in social stereotypes

    Charles M Judd and Bernadette Park. Definition and assessment of accuracy in social stereotypes. Psychological review, 100 0 (1): 0 109, 1993

    1993

  18. [26]

    Speech & language processing

    Dan Jurafsky. Speech & language processing. Pearson Education India, 2000

    2000

  19. [27]

    A perspective on judgment and choice: Mapping bounded rationality

    Daniel Kahneman. A perspective on judgment and choice: Mapping bounded rationality. Progress in Psychological Science around the World. Volume 1 Neural, Cognitive and Developmental Issues., pp.\ 1--47, 2013

    2013

  20. [28]

    Subjective probability: A judgment of representativeness

    Daniel Kahneman and Amos Tversky. Subjective probability: A judgment of representativeness. Cognitive psychology, 3 0 (3): 0 430--454, 1972

    1972

  21. [29]

    On the psychology of prediction

    Daniel Kahneman and Amos Tversky. On the psychology of prediction. Psychological review, 80 0 (4): 0 237, 1973

    1973

  22. [30]

    Alignment of language agents

    Zachary Kenton, Tom Everitt, Laura Weidinger, Iason Gabriel, Vladimir Mikulik, and Geoffrey Irving. Alignment of language agents. arXiv preprint arXiv:2103.14659, 2021

    2021 arXiv

  23. [31]

    The unlocking spell on base llms: Rethinking alignment via in-context learning

    Bill Yuchen Lin, Abhilasha Ravichander, Ximing Lu, Nouha Dziri, Melanie Sclar, Khyathi Chandu, Chandra Bhagavatula, and Yejin Choi. The unlocking spell on base llms: Rethinking alignment via in-context learning. arXiv preprint arXiv:2312.01552, 2023

    2023 arXiv

  24. [32]

    Xml retrieval

    Christopher D Manning, Prabhakar Raghavan, and Hinriche Sch \"u tze. Xml retrieval. Introduction to Information Retrieval, 2008

    2008

  25. [33]

    Evaluating cognitive maps and planning in large language models with cogeval

    Ida Momennejad, Hosein Hasanbeig, Felipe Vieira, Hiteshi Sharma, Robert Osazuwa Ness, Nebojsa Jojic, Hamid Palangi, and Jonathan Larson. Evaluating cognitive maps and planning in large language models with cogeval. arXiv preprint arXiv:2309.15129, 2023

    2023 arXiv

  26. [34]

    Stereoset: Measuring stereotypical bias in pretrained language models

    Moin Nadeem, Anna Bethke, and Siva Reddy. Stereoset: Measuring stereotypical bias in pretrained language models. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processin...

    2021

  27. [35]

    Spontaneous discounting of availability in frequency judgment tasks

    Daniel M Oppenheimer. Spontaneous discounting of availability in frequency judgment tasks. Psychological Science, 15 0 (2): 0 100--105, 2004

    2004

  28. [36]

    Training language models to follow instructions with human feedback

    Long Ouyang, Jeffrey Wu, Xu Jiang, Diogo Almeida, Carroll Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, et al. Training language models to follow instructions with human feedback. Advances in Neural Information Processing Systems, 35: 0 2...

    2022

  29. [37]

    Asian american and pacific islander students: Equity and the achievement gap

    Valerie Ooka Pang, Peggy P Han, and Jennifer M Pang. Asian american and pacific islander students: Equity and the achievement gap. Educational Researcher, 40 0 (8): 0 378--389, 2011

    2011

  30. [38]

    Identifying the risks of lm agents with an lm-emulated sandbox

    Yangjun Ruan, Honghua Dong, Andrew Wang, Silviu Pitis, Yongchao Zhou, Jimmy Ba, Yann Dubois, Chris J Maddison, and Tatsunori Hashimoto. Identifying the risks of lm agents with an lm-emulated sandbox. arXiv preprint arXiv:2309.15817, 2023

    2023 arXiv

  31. [39]

    Whose opinions do language models reflect? 2023

    Shibani Santurkar, Esin Durmus, Faisal Ladhak, Cinoo Lee, Percy Liang, and Tatsunori Hashimoto. Whose opinions do language models reflect? 2023

    2023

  32. [40]

    The psychology of stereotyping

    David J Schneider. The psychology of stereotyping. Guilford Press, 2005

    2005

  33. [41]

    Ease of retrieval as information: Another look at the availability heuristic

    Norbert Schwarz, Herbert Bless, Fritz Strack, Gisela Klumpp, Helga Rittenauer-Schatka, and Annette Simons. Ease of retrieval as information: Another look at the availability heuristic. Journal of Personality and Social psychology, 61 0 (2): 0 195, 1991

    1991

  34. [42]

    Towards understanding sycophancy in language models

    Mrinank Sharma, Meg Tong, Tomasz Korbak, David Duvenaud, Amanda Askell, Samuel R Bowman, Newton Cheng, Esin Durmus, Zac Hatfield-Dodds, Scott R Johnston, et al. Towards understanding sycophancy in language models. arXiv preprint arXiv:2310.13548, 2023

    arXiv

  35. [43]

    The trickle-down impact of reward (in-) consistency on rlhf

    Lingfeng Shen, Sihao Chen, Linfeng Song, Lifeng Jin, Baolin Peng, Haitao Mi, Daniel Khashabi, and Dong Yu. The trickle-down impact of reward (in-) consistency on rlhf. arXiv preprint arXiv:2309.16155, 2023

    2023 arXiv

  36. [44]

    Moral mimicry: Large language models produce moral rationalizations tailored to political identity

    Gabriel Simmons. Moral mimicry: Large language models produce moral rationalizations tailored to political identity. arXiv preprint arXiv:2209.12106, 2022

    2022 arXiv

  37. [45]

    Distributional preference learning: Understanding and accounting for hidden context in rlhf

    Anand Siththaranjan, Cassidy Laidlaw, and Dylan Hadfield-Menell. Distributional preference learning: Understanding and accounting for hidden context in rlhf. arXiv preprint arXiv:2312.08358, 2023

    2023 arXiv

  38. [46]

    Comparison of bayesian and regression approaches to the study of information processing in judgment

    Paul Slovic and Sarah Lichtenstein. Comparison of bayesian and regression approaches to the study of information processing in judgment. Organizational behavior and human performance, 6 0 (6): 0 649--744, 1971

    1971

  39. [47]

    ANES Time Series Cumulative Data File [dataset and documentation]

    American National Election Studies. ANES Time Series Cumulative Data File [dataset and documentation]. September 16, 2022 version , 2022. URL www.electionstudies.org

    2022

  40. [48]

    Deep alignment with country shrinks the moral gap between conservatives and liberals

    Sanaz Talaifar and William B Swann Jr. Deep alignment with country shrinks the moral gap between conservatives and liberals. Political Psychology, 40 0 (3): 0 657--675, 2019

    2019

  41. [49]

    Evaluating and mitigating discrimination 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 discrimination in language model decisions. arXiv preprint arXiv:2312.03689, 2023

    2023 arXiv

  42. [50]

    Gemini: a family of highly capable multimodal models

    Gemini Team, Rohan Anil, Sebastian Borgeaud, Yonghui Wu, Jean-Baptiste Alayrac, Jiahui Yu, Radu Soricut, Johan Schalkwyk, Andrew M Dai, Anja Hauth, et al. Gemini: a family of highly capable multimodal models. arXiv preprint arXiv:2312.11805, 2023

    2023 arXiv

  43. [51]

    Qwen2.5: A party of foundation models, September 2024

    Qwen Team. Qwen2.5: A party of foundation models, September 2024. URL https://qwenlm.github.io/blog/qwen2.5/

    2024

  44. [52]

    Llama 2: Open foundation and fine-tuned chat models

    Hugo Touvron, Louis Martin, Kevin Stone, Peter Albert, Amjad Almahairi, Yasmine Babaei, Nikolay Bashlykov, Soumya Batra, Prajjwal Bhargava, Shruti Bhosale, et al. Llama 2: Open foundation and fine-tuned chat models. arXiv preprint arXiv:2307.09288, 2023

    2023 arXiv

  45. [53]

    Judgment under uncertainty: Heuristics and biases: Biases in judgments reveal some heuristics of thinking under uncertainty

    Amos Tversky and Daniel Kahneman. Judgment under uncertainty: Heuristics and biases: Biases in judgments reveal some heuristics of thinking under uncertainty. science, 185 0 (4157): 0 1124--1131, 1974

    1974

  46. [54]

    liar, liar pants on fire

    William Yang Wang. “liar, liar pants on fire”: A new benchmark dataset for fake news detection. In Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), pp.\ 422--426, 2017

    2017

  47. [55]

    Chain-of-thought prompting elicits reasoning in large language models

    Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Fei Xia, Ed Chi, Quoc V Le, Denny Zhou, et al. Chain-of-thought prompting elicits reasoning in large language models. Advances in Neural Information Processing Systems, 35: 0 24824--24837, 2022

    2022

  48. [56]

    Qwen2 technical report

    An Yang, Baosong Yang, Binyuan Hui, Bo Zheng, Bowen Yu, Chang Zhou, Chengpeng Li, Chengyuan Li, Dayiheng Liu, Fei Huang, Guanting Dong, Haoran Wei, Huan Lin, Jialong Tang, Jialin Wang, Jian Yang, Jianhong Tu, Jianwei Zhang, Jianxin Ma, Jin Xu, Jingren Zhou, Jinze Bai, Jinzheng...

    2024 arXiv

  49. [57]

    From heuristic to analytic: Cognitively motivated strategies for coherent physical commonsense reasoning

    Zheyuan Zhang, Shane Storks, Fengyuan Hu, Sungryull Sohn, Moontae Lee, Honglak Lee, and Joyce Chai. From heuristic to analytic: Cognitively motivated strategies for coherent physical commonsense reasoning. In Houda Bouamor, Juan Pino, and Kalika Bali (eds.), Proceedings of the...

    2023 doi

  50. [58]

    Lima: Less is more for alignment

    Chunting Zhou, Pengfei Liu, Puxin Xu, Srini Iyer, Jiao Sun, Yuning Mao, Xuezhe Ma, Avia Efrat, Ping Yu, Lili Yu, et al. Lima: Less is more for alignment. arXiv preprint arXiv:2305.11206, 2023

    2023 arXiv

  51. [59]

    Efficiently measuring the cognitive ability of llms: An adaptive testing perspective

    Yan Zhuang, Qi Liu, Yuting Ning, Weizhe Huang, Rui Lv, Zhenya Huang, Guanhao Zhao, Zheng Zhang, Qingyang Mao, Shijin Wang, et al. Efficiently measuring the cognitive ability of llms: An adaptive testing perspective. arXiv preprint arXiv:2306.10512, 2023

    2023 arXiv

  52. [60]

    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 gl...

  53. [61]

    @esa (Ref

    \@ifxundefined[1] #1\@undefined \@firstoftwo \@secondoftwo \@ifnum[1] #1 \@firstoftwo \@secondoftwo \@ifx[1] #1 \@firstoftwo \@secondoftwo [2] @ #1 \@temptokena #2 #1 @ \@temptokena \@ifclassloaded agu2001 natbib The agu2001 class already includes natbib coding, so you should ...

  54. [62]

    \@lbibitem[] @bibitem@first@sw\@secondoftwo \@lbibitem[#1]#2 \@extra@b@citeb \@ifundefined br@#2\@extra@b@citeb \@namedef br@#2 \@nameuse br@#2\@extra@b@citeb \@ifundefined b@#2\@extra@b@citeb @num @parse #2 @tmp #1 NAT@b@open@#2 NAT@b@shut@#2 \@ifnum @merge>\@ne @bibitem@firs...

  55. [63]

    @open @close @open @close and [1] URL: #1 \@ifundefined chapter * \@mkboth \@ifxundefined @sectionbib * \@mkboth * \@mkboth\@gobbletwo \@ifclassloaded amsart * \@ifclassloaded amsbook * \@ifxundefined @heading @heading NAT@ctr thebibliography [1] @ \@biblabel @NAT@ctr \@bibset...

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.