@inproceedings{vargas-etal-2026-scalable,
title = "Scalable Multi-Agent {LLM}s for Multi-Turn Conversation Bias Detection and Mitigation",
author = "Vargas, Victor H. and
Haque, Farsheed and
Xu, Depeng and
Niu, Xi",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-emnlp.1314/",
doi = "10.18653/2026.findings-emnlp.1314",
pages = "27827--27843",
ISBN = "979-8-89176-462-0",
abstract = "Large language models (LLMs) increasingly participate in conversational systems where responses emerge through multi-turn interactions rather than single prompts. We introduce a multi-agent evaluation framework that probes LLMs for bias through multi-turn conversations. We present a framework that orchestrates three roles (target, adversary, and scorer) under diverse normative fairness definitions. Experiments on five widely used LLMs using 100 adversarial prompts show that adversarial multi-turn probing reliably surfaces bias-related responses, with baseline violation rates exceeding 80{\%} across models, while human validation with the scorer ({\textasciitilde}85{\%} agreement) confirms the scoring reliability. Preventive prompt-conditioning strategies provide limited reduction in violations, while our iterative post-hoc mitigation pipeline substantially reduces scorer-detected violations under the active fairness profile, from over 80{\%} to near-zero. We frame this as scorer-relative violation reduction rather than universal debiasing."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="vargas-etal-2026-scalable">
<titleInfo>
<title>Scalable Multi-Agent LLMs for Multi-Turn Conversation Bias Detection and Mitigation</title>
</titleInfo>
<name type="personal">
<namePart type="given">Victor</namePart>
<namePart type="given">H</namePart>
<namePart type="family">Vargas</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Farsheed</namePart>
<namePart type="family">Haque</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Depeng</namePart>
<namePart type="family">Xu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xi</namePart>
<namePart type="family">Niu</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>Large language models (LLMs) increasingly participate in conversational systems where responses emerge through multi-turn interactions rather than single prompts. We introduce a multi-agent evaluation framework that probes LLMs for bias through multi-turn conversations. We present a framework that orchestrates three roles (target, adversary, and scorer) under diverse normative fairness definitions. Experiments on five widely used LLMs using 100 adversarial prompts show that adversarial multi-turn probing reliably surfaces bias-related responses, with baseline violation rates exceeding 80% across models, while human validation with the scorer (~85% agreement) confirms the scoring reliability. Preventive prompt-conditioning strategies provide limited reduction in violations, while our iterative post-hoc mitigation pipeline substantially reduces scorer-detected violations under the active fairness profile, from over 80% to near-zero. We frame this as scorer-relative violation reduction rather than universal debiasing.</abstract>
<identifier type="citekey">vargas-etal-2026-scalable</identifier>
<identifier type="doi">10.18653/2026.findings-emnlp.1314</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.1314/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>27827</start>
<end>27843</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Scalable Multi-Agent LLMs for Multi-Turn Conversation Bias Detection and Mitigation
%A Vargas, Victor H.
%A Haque, Farsheed
%A Xu, Depeng
%A Niu, Xi
%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 vargas-etal-2026-scalable
%X Large language models (LLMs) increasingly participate in conversational systems where responses emerge through multi-turn interactions rather than single prompts. We introduce a multi-agent evaluation framework that probes LLMs for bias through multi-turn conversations. We present a framework that orchestrates three roles (target, adversary, and scorer) under diverse normative fairness definitions. Experiments on five widely used LLMs using 100 adversarial prompts show that adversarial multi-turn probing reliably surfaces bias-related responses, with baseline violation rates exceeding 80% across models, while human validation with the scorer (~85% agreement) confirms the scoring reliability. Preventive prompt-conditioning strategies provide limited reduction in violations, while our iterative post-hoc mitigation pipeline substantially reduces scorer-detected violations under the active fairness profile, from over 80% to near-zero. We frame this as scorer-relative violation reduction rather than universal debiasing.
%R 10.18653/2026.findings-emnlp.1314
%U https://aclanthology.org/2026.findings-emnlp.1314/
%U https://doi.org/10.18653/2026.findings-emnlp.1314
%P 27827-27843
Markdown (Informal)
[Scalable Multi-Agent LLMs for Multi-Turn Conversation Bias Detection and Mitigation](https://aclanthology.org/2026.findings-emnlp.1314/) (Vargas et al., Findings 2026)
ACL