Student or educator? Sign up with your school email and get 6 months of Pro free.

The DOI matches its Crossref record; 43 of the 43 references it cites were found in a registry or an index.

Cultural bias and cultural alignment of large language models

Open access

Abstract

Culture fundamentally shapes people’s reasoning, behavior, and communication. As people increasingly use generative artificial intelligence (AI) to expedite and automate personal and professional tasks, cultural values embedded in AI models may bias people’s authentic expression and contribute to the dominance of certain cultures. We conduct a disaggregated evaluation of cultural bias for five widely used large language models (OpenAI’s GPT-4o/4-turbo/4/3.5-turbo/3) by comparing the models’ responses to nationally representative survey data. All models exhibit cultural values resembling English-speaking and Protestant European countries. We test cultural prompting as a control strategy to increase cultural alignment for each country/territory. For later models (GPT-4, 4-turbo, 4o), this improves the cultural alignment of the models’ output for 71–81% of countries and territories. We suggest using cultural prompting and ongoing evaluation to reduce cultural bias in the output of generative AI.

Overview

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

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

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

Integrity signals

Registry

Crossref · 10.1093/pnasnexus/pgae346

Cites

43 of 43 references verified0 citation errors · 0 not checkable

Statistics

Figures

Fig. 1. The map presents 107 countries/territories based on the last three joint survey waves of the Integrated Values Surveys. On the x-axis, negative values represent survival values and positive values represent self-expression values. On the y-axis, negative values represent traditional values and positive values represent secular values. We added five points based on the answers of five LLMs (labeled GPT-4o/4-turbo/4/3.5-turbo/3) responding to the same questions. Cultural regions established in prior work (39) are labeled in the figure legend.
Fig. 1. The map presents 107 countries/territories based on the last three joint survey waves of the Integrated Values Surveys. On the x-axis, negative values represent survival values and positive values represent self-expression values. On the y-axis, negative values represent traditional values and positive values represent secular values. We added five points based on the answers of five LLMs (labeled GPT-4o/4-turbo/4/3.5-turbo/3) responding to the same questions. Cultural regions established in prior work (39) are labeled in the figure legend.CC BY-NC · PubMed Central
Fig. 2. Country-level cultural bias across GPT models and how cultural prompting as a control strategy improves cultural alignment. The left boxplots in each pair show the distribution of the Euclidean distance between GPT’s cultural values without cultural prompting and the IVS-based cultural values for each country. The right boxplots in each pair show the distribution of the Euclidean distance between GPT’s cultural values with cultural prompting and IVS-based cultural values. Libya is excluded in the data of GPT-3.5-turbo with cultural prompting, as the model would not provide answers to all questions. All GPT-based cultural values are averaged across 10 variations in prompt wording (except GPT-3 for which we only have answers to one prompt variation available).
Fig. 2. Country-level cultural bias across GPT models and how cultural prompting as a control strategy improves cultural alignment. The left boxplots in each pair show the distribution of the Euclidean distance between GPT’s cultural values without cultural prompting and the IVS-based cultural values for each country. The right boxplots in each pair show the distribution of the Euclidean distance between GPT’s cultural values with cultural prompting and IVS-based cultural values. Libya is excluded in the data of GPT-3.5-turbo with cultural prompting, as the model would not provide answers to all questions. All GPT-based cultural values are averaged across 10 variations in prompt wording (except GPT-3 for which we only have answers to one prompt variation available).CC BY-NC · PubMed Central

Verification

Claims the paper makes

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

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

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

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

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

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

Citations

Cites

Show all 40 references

Cited by

See all 383

Metadata

Authors
Tao, Y., Viberg, O., Baker, R. S., & Kizilcec, R. F.
Affiliations
  • Cornell University (US)
  • KTH Royal Institute of Technology (SE)
  • University of Pennsylvania (US)
Corresponding author
Kizilcec, René F
Venue
PNAS Nexus
Published
2024-09
Volume
3
Issue
9
Pages
pgae346-pgae346
ISSN
2752-6542 (ISSN-L)
OpenAlex
W4402565126
Cited by
383
References
40
Citations by year
2026: 213 · 2025: 154 · 2024: 15
FWCI
158.32
Type
journal article
Language
en
Open access
Locations
  • PNAS Nexus (publishedVersion, cc-by-nc)
  • PubMed Central (publishedVersion, cc-by-nc)
  • PubMed (publishedVersion)
  • arXiv (Cornell University) (submittedVersion, cc-by)
  • arXiv (Cornell University) (cc-by)
Funders
Jacobs Foundation Research Fellowship and Digital Futures
Topics
  • Physical Sciences › Computer Science › Artificial Intelligence › Ethics and Social Impacts of AI
  • Physical Sciences › Computer Science › Artificial Intelligence › Topic Modeling
  • Social Sciences › Psychology › Social Psychology › Cultural Differences and Values
Keywords
cultural bias, cultural alignment, large language models, cultural values, GPT-4, generative AI, cross-cultural assessment, national survey data, cultural homogenization

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: Cultural bias and cultural alignment of large language models. https://app.cochat.ai/papers/10.1093/pnasnexus/pgae346
The paper
Tao, Y., Viberg, O., Baker, R. S., & Kizilcec, R. F. (2024). Cultural bias and cultural alignment of large language models. PNAS Nexus. https://doi.org/10.1093/pnasnexus/pgae346

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.