@inproceedings{sakhawat-etal-2026-aidg,
title = "{AIDG}: A Formal Decomposition of Information Extraction and Containment Asymmetries in Multi-Turn {LLM} Dialogue",
author = "Sakhawat, Adib and
Sadab, Fardeen and
Shahriar, Rakin",
editor = "Kamalloo, Ehsan and
Gontier, Nicolas and
Rozanov, Nikolai and
Hashemzadeh, Maryam and
Arora, Akhil",
booktitle = "Proceedings of the Second Workshop for {RE}search on Agent Language Models ({REALM} 2026)",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.realm-1.40/",
pages = "579--598",
ISBN = "979-8-89176-485-9",
abstract = "Multi-turn LLM evaluation is typically reported as a single win-rate scalar, conflating distinct capabilities. We introduce AIDG (Adversarial Information Deduction Game), formalizing multi-turn adversarial dialogue as a two-player partially observable stochastic game (POSG) and decomposing performance along Seeker (extraction) and Holder (containment) roles. The decomposition isolates three failure modes: cooperative-prior leakage, constraint{--}reasoning interference, and inefficient hypothesis-space traversal. Across 439 games over six frontier LLMs, defensive performance is tightly clustered ({\ensuremath{\sigma}} = 1.9 ELO) while offensive performance varies substantially ({\ensuremath{\sigma}} = 53.3 ELO); confirmation framing increases extraction odds 7.75{\texttimes} over uninformed deduction (p {\ensuremath{<}} 0.00001); and constraint violations account for 41.3{\%} of deductive failures, uncorrelated with scale ({\ensuremath{\rho}} = 0.0). We position the containment-over-extraction gap not as a surprising finding but as a measurable consequence of locally-resolvable defensive decisions versus globally-coupled offensive planning, and use the decomposition to attribute the gap per model. All design choices, including turn-decay weighting and the Bradley{--}Terry rating model, are derived from explicit assumptions."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="sakhawat-etal-2026-aidg">
<titleInfo>
<title>AIDG: A Formal Decomposition of Information Extraction and Containment Asymmetries in Multi-Turn LLM Dialogue</title>
</titleInfo>
<name type="personal">
<namePart type="given">Adib</namePart>
<namePart type="family">Sakhawat</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Fardeen</namePart>
<namePart type="family">Sadab</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Rakin</namePart>
<namePart type="family">Shahriar</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 Second Workshop for REsearch on Agent Language Models (REALM 2026)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Ehsan</namePart>
<namePart type="family">Kamalloo</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Nicolas</namePart>
<namePart type="family">Gontier</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Nikolai</namePart>
<namePart type="family">Rozanov</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Maryam</namePart>
<namePart type="family">Hashemzadeh</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Akhil</namePart>
<namePart type="family">Arora</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">979-8-89176-485-9</identifier>
</relatedItem>
<abstract>Multi-turn LLM evaluation is typically reported as a single win-rate scalar, conflating distinct capabilities. We introduce AIDG (Adversarial Information Deduction Game), formalizing multi-turn adversarial dialogue as a two-player partially observable stochastic game (POSG) and decomposing performance along Seeker (extraction) and Holder (containment) roles. The decomposition isolates three failure modes: cooperative-prior leakage, constraint–reasoning interference, and inefficient hypothesis-space traversal. Across 439 games over six frontier LLMs, defensive performance is tightly clustered (\ensuremathσ = 1.9 ELO) while offensive performance varies substantially (\ensuremathσ = 53.3 ELO); confirmation framing increases extraction odds 7.75× over uninformed deduction (p \ensuremath< 0.00001); and constraint violations account for 41.3% of deductive failures, uncorrelated with scale (\ensuremathρ = 0.0). We position the containment-over-extraction gap not as a surprising finding but as a measurable consequence of locally-resolvable defensive decisions versus globally-coupled offensive planning, and use the decomposition to attribute the gap per model. All design choices, including turn-decay weighting and the Bradley–Terry rating model, are derived from explicit assumptions.</abstract>
<identifier type="citekey">sakhawat-etal-2026-aidg</identifier>
<location>
<url>https://aclanthology.org/2026.realm-1.40/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>579</start>
<end>598</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T AIDG: A Formal Decomposition of Information Extraction and Containment Asymmetries in Multi-Turn LLM Dialogue
%A Sakhawat, Adib
%A Sadab, Fardeen
%A Shahriar, Rakin
%Y Kamalloo, Ehsan
%Y Gontier, Nicolas
%Y Rozanov, Nikolai
%Y Hashemzadeh, Maryam
%Y Arora, Akhil
%S Proceedings of the Second Workshop for REsearch on Agent Language Models (REALM 2026)
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-485-9
%F sakhawat-etal-2026-aidg
%X Multi-turn LLM evaluation is typically reported as a single win-rate scalar, conflating distinct capabilities. We introduce AIDG (Adversarial Information Deduction Game), formalizing multi-turn adversarial dialogue as a two-player partially observable stochastic game (POSG) and decomposing performance along Seeker (extraction) and Holder (containment) roles. The decomposition isolates three failure modes: cooperative-prior leakage, constraint–reasoning interference, and inefficient hypothesis-space traversal. Across 439 games over six frontier LLMs, defensive performance is tightly clustered (\ensuremathσ = 1.9 ELO) while offensive performance varies substantially (\ensuremathσ = 53.3 ELO); confirmation framing increases extraction odds 7.75× over uninformed deduction (p \ensuremath< 0.00001); and constraint violations account for 41.3% of deductive failures, uncorrelated with scale (\ensuremathρ = 0.0). We position the containment-over-extraction gap not as a surprising finding but as a measurable consequence of locally-resolvable defensive decisions versus globally-coupled offensive planning, and use the decomposition to attribute the gap per model. All design choices, including turn-decay weighting and the Bradley–Terry rating model, are derived from explicit assumptions.
%U https://aclanthology.org/2026.realm-1.40/
%P 579-598
Markdown (Informal)
[AIDG: A Formal Decomposition of Information Extraction and Containment Asymmetries in Multi-Turn LLM Dialogue](https://aclanthology.org/2026.realm-1.40/) (Sakhawat et al., REALM 2026)
ACL