<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="wang-etal-2026-evasive">
    <titleInfo>
        <title>From Evasive Servants to Dignified Peers via Dual Preference Optimization</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Xiangqi</namePart>
        <namePart type="family">Wang</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Yue</namePart>
        <namePart type="family">Huang</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Haomin</namePart>
        <namePart type="family">Zhuang</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Kehan</namePart>
        <namePart type="family">Guo</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Xiangliang</namePart>
        <namePart type="family">Zhang</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>Findings of the Association for Computational Linguistics: EMNLP 2026</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">979-8-89176-462-0</identifier>
    </relatedItem>
    <abstract>Current aligned language models exhibit a dual failure mode we term the Evasive Servant: they sycophantically validate flawed user beliefs while deflecting responsibility with boilerplate disclaimers. We propose the Dignified Peer framework, which targets servility through anti-sycophancy and trustworthiness, and evasiveness through empathy and creativity. Operationalizing this framework requires addressing three concrete gaps: lack of fine-grained per-dimension supervision, objective imbalance during multi-objective optimization, and confound-prone evaluation. We introduce the PersonaKnob dataset, which encodes a compositional partial-order structure of multi-persona preferences, and pair it with Lag-DPO, a tolerance-constrained Lagrangian DPO algorithm that dynamically rebalances dimensions to mitigate collapse. For evaluation, we adapt the Many-Facet Rasch Model (MFRM) to disentangle model capability from judge severity, rubric difficulty, and question complexity, and we validate MFRM scores against human pairwise preferences (Spearman ρ ın [0.67, 0.85] across dimensions). Across two open-source backbones, Lag-DPO improves all four dimensions in distribution and reduces sycophancy while preserving safe-refusal balance on out-of-distribution benchmarks; we report these results with explicit weight-sensitivity and uncertainty analyses rather than claim a fully resolved Dignified Peer.</abstract>
    <identifier type="citekey">wang-etal-2026-evasive</identifier>
    <location>
        <url>https://aclanthology.org/2026.findings-emnlp.197/</url>
    </location>
    <part>
        <date>2026-10</date>
        <extent unit="page">
            <start>4164</start>
            <end>4186</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T From Evasive Servants to Dignified Peers via Dual Preference Optimization
%A Wang, Xiangqi
%A Huang, Yue
%A Zhuang, Haomin
%A Guo, Kehan
%A Zhang, Xiangliang
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Findings of the Association for Computational Linguistics: EMNLP 2026
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-462-0
%F wang-etal-2026-evasive
%X Current aligned language models exhibit a dual failure mode we term the Evasive Servant: they sycophantically validate flawed user beliefs while deflecting responsibility with boilerplate disclaimers. We propose the Dignified Peer framework, which targets servility through anti-sycophancy and trustworthiness, and evasiveness through empathy and creativity. Operationalizing this framework requires addressing three concrete gaps: lack of fine-grained per-dimension supervision, objective imbalance during multi-objective optimization, and confound-prone evaluation. We introduce the PersonaKnob dataset, which encodes a compositional partial-order structure of multi-persona preferences, and pair it with Lag-DPO, a tolerance-constrained Lagrangian DPO algorithm that dynamically rebalances dimensions to mitigate collapse. For evaluation, we adapt the Many-Facet Rasch Model (MFRM) to disentangle model capability from judge severity, rubric difficulty, and question complexity, and we validate MFRM scores against human pairwise preferences (Spearman ρ ın [0.67, 0.85] across dimensions). Across two open-source backbones, Lag-DPO improves all four dimensions in distribution and reduces sycophancy while preserving safe-refusal balance on out-of-distribution benchmarks; we report these results with explicit weight-sensitivity and uncertainty analyses rather than claim a fully resolved Dignified Peer.
%U https://aclanthology.org/2026.findings-emnlp.197/
%P 4164-4186

Markdown (Informal)

[From Evasive Servants to Dignified Peers via Dual Preference Optimization](https://aclanthology.org/2026.findings-emnlp.197/) (Wang et al., Findings 2026)

• From Evasive Servants to Dignified Peers via Dual Preference Optimization (Wang et al., Findings 2026)

ACL

• Xiangqi Wang, Yue Huang, Haomin Zhuang, Kehan Guo, and Xiangliang Zhang. 2026. From Evasive Servants to Dignified Peers via Dual Preference Optimization. In Findings of the Association for Computational Linguistics: EMNLP 2026, pages 4164–4186, Budapest, Hungary. Association for Computational Linguistics.