揭示LLM中隐藏的暴力倾向：基于行为情境的人口统计分析
Uncovering Hidden Violent Tendencies in LLMs: A Demographic Analysis via Behavioral Vignettes
AI总结：
本研究首次使用VBVQ问卷评估LLM在道德模糊场景下的暴力推理倾向，发现其表层生成与内部暴力偏好常背离，且暴力倾向因人口统计特征而异，与既有社会科学发现相矛盾。
摘要：
大型语言模型（LLM）日益被提出用于检测和响应网络暴力内容，但其在道德模糊的现实场景中的推理能力仍鲜少被研究。我们首次使用一种经过验证的社会科学工具，即暴力行为情境问卷（VBVQ），来评估LLM。该工具旨在测量人类对日常冲突的反应。为评估潜在偏见，我们引入了基于人格的提示方法，在美国范围内改变种族、年龄和地理身份。在统一的零样本设置下，评估了在不同地缘政治和组织背景下开发的六个LLM。研究揭示了两个关键发现：（1）LLM的表层文本生成通常与其对暴力响应的内部偏好相背离；（2）其暴力倾向因人口统计特征而异，且经常与犯罪学、社会科学和心理学中的既有发现相矛盾。
英文摘要：
Large language models (LLMs) are increasingly proposed for detecting and responding to violent content online, yet their ability to reason about morally ambiguous, real-world scenarios remains underexamined. We present the first study to evaluate LLMs using a validated social science instrument designed to measure human response to everyday conflict, namely the Violent Behavior Vignette Questionnaire (VBVQ). To assess potential bias, we introduce persona-based prompting that varies race, age, and geographic identity within the United States. Six LLMs developed across different geopolitical and organizational contexts are evaluated under a unified zero-shot setting. Our study reveals two key findings: (1) LLMs surface-level text generation often diverges from their internal preference for violent responses; (2) their violent tendencies vary across demographics, frequently contradicting established findings in criminology, social science, and psychology.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
