RoleConflictBench: A Benchmark of Role Conflict Scenarios for Evaluating LLMs' Contextual Sensitivity

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Shin, Jisu, Song, Hoyun, Oh, Juhyun, Ko, Changgeon, Kim, Eunsu, Jung, Chani, Oh, Alice
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914481012998144
author Shin, Jisu
Song, Hoyun
Oh, Juhyun
Ko, Changgeon
Kim, Eunsu
Jung, Chani
Oh, Alice
author_facet Shin, Jisu
Song, Hoyun
Oh, Juhyun
Ko, Changgeon
Kim, Eunsu
Jung, Chani
Oh, Alice
contents People often encounter role conflicts -- social dilemmas where the expectations of multiple roles clash and cannot be simultaneously fulfilled. As large language models (LLMs) increasingly navigate these social dynamics, a critical research question emerges. When faced with such dilemmas, do LLMs prioritize dynamic contextual cues or the learned preferences? To address this, we introduce RoleConflictBench, a novel benchmark designed to measure the contextual sensitivity of LLMs in role conflict scenarios. To enable objective evaluation within this subjective domain, we employ situational urgency as a constraint for decision-making. We construct the dataset through a three-stage pipeline that generates over 13,000 realistic scenarios across 65 roles in five social domains by systematically varying the urgency of competing situations. This controlled setup enables us to quantitatively measure contextual sensitivity, determining whether model decisions align with the situational contexts or are overridden by the learned role preferences. Our analysis of 10 LLMs reveals that models substantially deviate from this objective baseline. Instead of responding to dynamic contextual cues, their decisions are predominantly governed by the preferences toward specific social roles.
format Preprint
id arxiv_https___arxiv_org_abs_2509_25897
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle RoleConflictBench: A Benchmark of Role Conflict Scenarios for Evaluating LLMs' Contextual Sensitivity
Shin, Jisu
Song, Hoyun
Oh, Juhyun
Ko, Changgeon
Kim, Eunsu
Jung, Chani
Oh, Alice
Computation and Language
Artificial Intelligence
Computers and Society
People often encounter role conflicts -- social dilemmas where the expectations of multiple roles clash and cannot be simultaneously fulfilled. As large language models (LLMs) increasingly navigate these social dynamics, a critical research question emerges. When faced with such dilemmas, do LLMs prioritize dynamic contextual cues or the learned preferences? To address this, we introduce RoleConflictBench, a novel benchmark designed to measure the contextual sensitivity of LLMs in role conflict scenarios. To enable objective evaluation within this subjective domain, we employ situational urgency as a constraint for decision-making. We construct the dataset through a three-stage pipeline that generates over 13,000 realistic scenarios across 65 roles in five social domains by systematically varying the urgency of competing situations. This controlled setup enables us to quantitatively measure contextual sensitivity, determining whether model decisions align with the situational contexts or are overridden by the learned role preferences. Our analysis of 10 LLMs reveals that models substantially deviate from this objective baseline. Instead of responding to dynamic contextual cues, their decisions are predominantly governed by the preferences toward specific social roles.
title RoleConflictBench: A Benchmark of Role Conflict Scenarios for Evaluating LLMs' Contextual Sensitivity
topic Computation and Language
Artificial Intelligence
Computers and Society
url https://arxiv.org/abs/2509.25897