<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="ram-etal-2026-consensus">
    <titleInfo>
        <title>Consensus Drift: Semantic Salience Can Override Majority Evidence in LLM Aggregation</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Gokul</namePart>
        <namePart type="given">Srinath</namePart>
        <namePart type="given">Seetha</namePart>
        <namePart type="family">Ram</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Rashmi</namePart>
        <namePart type="family">Elavazhagan</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Anirudh</namePart>
        <namePart type="family">Vasudevan</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2026-10</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <relatedItem type="host">
        <titleInfo>
            <title>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing</title>
        </titleInfo>
        <name type="personal">
            <namePart type="given">Hung-Yi</namePart>
            <namePart type="family">Lee</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Sunipa</namePart>
            <namePart type="family">Dev</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Jessy</namePart>
            <namePart type="family">Li</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Desmond</namePart>
            <namePart type="family">Elliott</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <originInfo>
            <publisher>Association for Computational Linguistics</publisher>
            <place>
                <placeTerm type="text">Budapest, Hungary</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
        <identifier type="isbn">979-8-89176-459-0</identifier>
    </relatedItem>
    <abstract>We study how language models aggregate conflicting evidence in controlled multi-juror settings. As a null control, symbolic majority voting, where jurors provide only YES/NO votes, yields near-zero minority override across models, confirming this is not an artifact of the prompt family. Under semantic aggregation, where jurors instead provide short evidence statements paired with local verdicts, override becomes strength-dependent and position-sensitive: rates rise when a minority statement is operationalized as stronger than a weak majority, while a strong-majority/weak-minority control remains low-override. Authority cues applied to a minority statement can modulate this effect in some contexts, but do not universally dominate evidence strength. We validate these patterns with seed replications, cross-domain templates, phase-transition mapping across majority strength, minority strength, and position, and rationality-vs-salience controls. We further confirm the pattern on a larger open-weight model (Llama-3.3-70B-Instruct) and the GPT-4.1 flagship model: Llama-3.3-70B stays close to the smaller-model baseline, while GPT-4.1 shows a substantially larger override magnitude, indicating the effect’s size is model-dependent even as its qualitative shape persists. We emphasize that evidence “strength” here is a synthetic, operationalized manipulation rather than real-world ground truth, and that minority-following can be rational when minority evidence is genuinely decisive.</abstract>
    <identifier type="citekey">ram-etal-2026-consensus</identifier>
    <location>
        <url>https://aclanthology.org/2026.emnlp-main.1689/</url>
    </location>
    <part>
        <date>2026-10</date>
        <extent unit="page">
            <start>38411</start>
            <end>38428</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T Consensus Drift: Semantic Salience Can Override Majority Evidence in LLM Aggregation
%A Ram, Gokul Srinath Seetha
%A Elavazhagan, Rashmi
%A Vasudevan, Anirudh
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F ram-etal-2026-consensus
%X We study how language models aggregate conflicting evidence in controlled multi-juror settings. As a null control, symbolic majority voting, where jurors provide only YES/NO votes, yields near-zero minority override across models, confirming this is not an artifact of the prompt family. Under semantic aggregation, where jurors instead provide short evidence statements paired with local verdicts, override becomes strength-dependent and position-sensitive: rates rise when a minority statement is operationalized as stronger than a weak majority, while a strong-majority/weak-minority control remains low-override. Authority cues applied to a minority statement can modulate this effect in some contexts, but do not universally dominate evidence strength. We validate these patterns with seed replications, cross-domain templates, phase-transition mapping across majority strength, minority strength, and position, and rationality-vs-salience controls. We further confirm the pattern on a larger open-weight model (Llama-3.3-70B-Instruct) and the GPT-4.1 flagship model: Llama-3.3-70B stays close to the smaller-model baseline, while GPT-4.1 shows a substantially larger override magnitude, indicating the effect’s size is model-dependent even as its qualitative shape persists. We emphasize that evidence “strength” here is a synthetic, operationalized manipulation rather than real-world ground truth, and that minority-following can be rational when minority evidence is genuinely decisive.
%U https://aclanthology.org/2026.emnlp-main.1689/
%P 38411-38428

Markdown (Informal)

[Consensus Drift: Semantic Salience Can Override Majority Evidence in LLM Aggregation](https://aclanthology.org/2026.emnlp-main.1689/) (Ram et al., EMNLP 2026)

• Consensus Drift: Semantic Salience Can Override Majority Evidence in LLM Aggregation (Ram et al., EMNLP 2026)

ACL

• Gokul Srinath Seetha Ram, Rashmi Elavazhagan, and Anirudh Vasudevan. 2026. Consensus Drift: Semantic Salience Can Override Majority Evidence in LLM Aggregation. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 38411–38428, Budapest, Hungary. Association for Computational Linguistics.