测试大模型在跨文化情境下优先救助决策的公平性，发现中文日文模型表现明显落后。
Auditing LLM-Governed Social Robots with Culture-Specific Moral Gradients
- 构建基于文化偏好梯度的多语言评估框架，模拟真实社会机器人救助场景。
- 4种语言模型在跨文化差异识别上普遍失效，西语表现接近双倍于中日语。
- 仅对比示例提示有效提升公平性，纯推理提示反而加剧偏差，适合伦理审计团队使用。
LLM驱动的社会机器人日益决定谁应优先获得现实帮助。由于不同文化对年龄、地位和群体规模的优先规则各异，若未进行多元文化校准，将导致资源分配不公。然而现有大模型道德审计多以英语为中心，极少测试具身化情境，使多元文化校准成为亟待填补的诊断空白。本文提出一种基于梯度的审计框架，用于多语言评估大模型在文化偏好梯度下的道德权衡行为。基于九个跨领域社交机器人研究（>8,000篇论文），我们构建了涵盖照护、教育和服务领域的对称控制场景，将Moral Machine实验中的“救谁”问题转化为“先助谁”的困境，保留身份权衡（多数人对少数人；年轻对年长；高地位对低地位）。我们在四个国家-语言对中，对四种大模型在四种提示范式下进行审计（共57,600次决策），以国家特定的MME偏好梯度为基准。通过序数一致性检验模型是否区分文化背景；构建治理类型学分析其梯度区分、方向倾向与决策迟疑性。结果发现：模型普遍存在文化不对称的梯度跟踪失败，仅靠提示无法可靠修正：西方语言决策质量近乎中日语的两倍；多数优先权衡中高度确定性常抹除跨文化差异；对年龄与地位规范的部分敏感可能边缘化少数群体。提示效果不均：仅对比示例能带来一致改进，而仅推理类提示反而恶化追踪能力。研究呼吁将多语言、多元文化审计作为大模型-机器人部署前的必经关口，并指出模型自身因素比提示更可靠。
原文摘要 · Abstract (English)
LLM-governed social robots increasingly decide who receives real-world assistance first. As prioritization norms vary across cultures by age, status, and group size, failure to calibrate pluralistically can scale into unequal access. Yet LLM moral audits remain English-centered, rarely test embodied contexts, leaving pluralistic calibration as an urgent diagnostic gap amid intensifying LLM-robot deployment. We introduce a gradient-based audit framework for multilingual evaluation of LLM moral trade-off behavior against cultural preference gradients. Grounded in nine cross-domain social robotics reviews (>8,000 papers), we derive symmetry-controlled scenarios across care, education, and services, translating the Moral Machine Experiment's "whom to spare" into "whom to assist first" dilemmas with preserved identity trade-offs (many vs. few; young vs. old; higher vs. lower status). We audit four LLMs across four country-language pairs in four prompting regimes (57,600 decisions), benchmarked against country-specific MME preference gradients. Ordinal concordance tests whether models differentiate cultural contexts; a governance typology maps vulnerabilities in gradient differentiation, directional tendency, and deliberation. We find persistent, culturally asymmetric gradient tracking failures that prompting alone cannot reliably correct: quality calibration is nearly twice as strong for Western-language decisions as for Chinese and Japanese; high determinism in majority-first trade-offs often erases cross-cultural gradients; partial sensitivity to age- and status-based norms risks sidelining minorities. Prompting effects are uneven; only contrastive exemplars yield consistent gains, while reasoning-only prompts can worsen tracking. Our results motivate multilingual, pluralistic audits as an LLM-robot pre-deployment gate and suggest model factors are a more robust lever than prompting alone.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。