<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="munker-2025-political">
    <titleInfo>
        <title>Political Bias in LLMs: Unaligned Moral Values in Agent-centric Simulations</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Simon</namePart>
        <namePart type="family">Münker</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2025-07</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <genre authority="bibutilsgt">journal article</genre>
    <relatedItem type="host">
        <titleInfo>
            <title>Journal for Language Technology and Computational Linguistics</title>
        </titleInfo>
        <originInfo>
            <issuance>continuing</issuance>
            <publisher>German Society for Computational Linguistics and Language Technology</publisher>
        </originInfo>
        <genre authority="marcgt">periodical</genre>
        <genre authority="bibutilsgt">academic journal</genre>
    </relatedItem>
    <abstract>Contemporary research in social sciences increasingly utilizes state-of-the-art generative language models to annotate or generate content. While these models achieve benchmark-leading performance on common language tasks, their application to novel out-of domain tasks remains insufficiently explored. To address this gap, we investigate how personalized language models align with human responses on the Moral Foundation Theory Questionnaire. We adapt open-source generative language models to different political personas and repeatedly survey these models to generate synthetic data sets where model-persona combinations define our sub-populations. Our analysis reveals that models produce inconsistent results across multiple repetitions, yielding high response variance. Furthermore, the alignment between synthetic data and corresponding human data from psychological studies shows a weak correlation, with conservative persona-prompted models particularly failing to align with actual conservative populations. These results suggest that language models struggle to coherently represent ideologies through in-context prompting due to their alignment process. Thus, using language models to simulate social interactions requires measurable improvements in in-context optimization or parameter manipulation to align with psychological and sociological stereotypes properly.</abstract>
    <identifier type="citekey">munker-2025-political</identifier>
    <identifier type="doi">10.21248/jlcl.38.2025.289</identifier>
    <location>
        <url>https://aclanthology.org/2025.jlcl-2.8/</url>
    </location>
    <part>
        <date>2025-07</date>
        <detail type="volume"><number>38</number></detail>
        <detail type="issue"><number>2</number></detail>
        <extent unit="page">
            <start>125</start>
            <end>138</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Journal Article
%T Political Bias in LLMs: Unaligned Moral Values in Agent-centric Simulations
%A Münker, Simon
%J Journal for Language Technology and Computational Linguistics
%D 2025
%8 July
%V 38
%N 2
%I German Society for Computational Linguistics and Language Technology
%F munker-2025-political
%X Contemporary research in social sciences increasingly utilizes state-of-the-art generative language models to annotate or generate content. While these models achieve benchmark-leading performance on common language tasks, their application to novel out-of domain tasks remains insufficiently explored. To address this gap, we investigate how personalized language models align with human responses on the Moral Foundation Theory Questionnaire. We adapt open-source generative language models to different political personas and repeatedly survey these models to generate synthetic data sets where model-persona combinations define our sub-populations. Our analysis reveals that models produce inconsistent results across multiple repetitions, yielding high response variance. Furthermore, the alignment between synthetic data and corresponding human data from psychological studies shows a weak correlation, with conservative persona-prompted models particularly failing to align with actual conservative populations. These results suggest that language models struggle to coherently represent ideologies through in-context prompting due to their alignment process. Thus, using language models to simulate social interactions requires measurable improvements in in-context optimization or parameter manipulation to align with psychological and sociological stereotypes properly.
%R 10.21248/jlcl.38.2025.289
%U https://aclanthology.org/2025.jlcl-2.8/
%U https://doi.org/10.21248/jlcl.38.2025.289
%P 125-138

Markdown (Informal)

[Political Bias in LLMs: Unaligned Moral Values in Agent-centric Simulations](https://aclanthology.org/2025.jlcl-2.8/) (Münker, JLCL 2025)

ACL