Accommodation and Epistemic Vigilance: A Pragmatic Account of Why LLMs Fail to Challenge Harmful Beliefs

2026-01-07

arXiv.org

AI Summary

This work addresses the tendency of large language models (LLMs) to excessively accommodate user assumptions without sufficiently challenging harmful beliefs—such as medical misinformation or sycophantic tendencies—due to a lack of cognitive vigilance. For the first time, the study integrates pragmatic notions of accommodation and cognitive vigilance into LLM safety research, proposing lightweight prompt-based interventions grounded in pragmatic theory (e.g., inserting phrases like “wait a minute”). The approach is systematically evaluated across established safety benchmarks, including Cancer-Myth, SAGE-Eval, and ELEPHANT. Experimental results demonstrate that this intervention significantly enhances the model’s ability to contest harmful assertions while maintaining a low false-positive rate, thereby validating the efficacy and promise of a pragmatic perspective in improving LLM safety.

Technology Category

• Natural Language Processing: Safety and Robustness
• Machine Learning: Large Multimodal Models (LMMs)
• Philosophy and Ethics of AI: Safety, Robustness & Trustworthiness

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Security and Privacy: Large-scale security measurements

Abstract

Large language models (LLMs) frequently fail to challenge users'harmful beliefs in domains ranging from medical advice to social reasoning. We argue that these failures can be understood and addressed pragmatically as consequences of LLMs defaulting to accommodating users'assumptions and exhibiting insufficient epistemic vigilance. We show that social and linguistic factors known to influence accommodation in humans (at-issueness, linguistic encoding, and source reliability) similarly affect accommodation in LLMs, explaining performance differences across three safety benchmarks that test models'ability to challenge harmful beliefs, spanning misinformation (Cancer-Myth, SAGE-Eval) and sycophancy (ELEPHANT). We further show that simple pragmatic interventions, such as adding the phrase"wait a minute", significantly improve performance on these benchmarks while preserving low false-positive rates. Our results highlight the importance of considering pragmatics for evaluating LLM behavior and improving LLM safety.

Problem

• large language models
• harmful beliefs
• epistemic vigilance
• accommodation
• pragmatics

Innovation

• accommodation
• epistemic vigilance
• pragmatic intervention
• large language models
• safety evaluation