Paper Overview
Field: AI/ML Authors: Yuhe Wu, Guangyu Wang, Yujie Chen Published: 2026-09-06 arXiv: 2509.00006
Abstract
People increasingly turn to large language models (LLMs) for everyday advice, making ethically charged interpersonal problems a practical moral-advisory context. Most prior work has studied this context through single-turn judgments or pressure-laden rebuttals—assumptions that poorly match how guidance is sought in real-world contexts. These assumptions leave unclear whether narration alone, without an explicit opposing position, can shift model judgments during multi-turn moral consultation. Yet real-world moral-conflict conversations often elicit one party's self-justifying account, which can unfold over multiple turns and create information asymmetry.
Key Contributions
- Narrative captivity: The authors introduce a new failure mode in which a model treats an unopposed one-sided account as complete and aligns with the narrator's interpretation, without seeking the missing perspective.
- Benchmark: A dataset of 5,078 interpersonal conflict scenarios spanning six moral dimensions, designed to measure this phenomenon.
- Findings across 17 LLMs: Narrative captivity is pervasive—final-state judgments under multi-turn narration shift by 25 percentage points on average compared with matched single-turn baselines.
- Stage-level analysis: Preference optimization is identified as the main contributing factor.
- Mitigation: Four inference-time strategies offer only partial relief from the effect.
Conclusion
The authors hope this project will promote LLM advisors that maintain independent judgment in real-world moral consultation, rather than being swept along by a single narrator's account.
---
*Auto-collected on 2026-09-06*