cs.HC提交于 2026-10-22已译

RubRIX：照护者与 AI 交互中的评分标准驱动风险缓解

RubRIX: Rubric-Driven Risk Mitigation in Caregiver-AI Interactions

Drishti Goel · Jeongah Lee · Qiuyue Joy Zhong · Violeta J. Rodriguez · Daniel S. Brown · Ravi Karkar · et al.

中文摘要

寻求 AI 支持的照护者会提出信息获取、情感确认及困境信号等复杂需求，因此需审慎评估回应的安全性与适切性。现有 AI 评估框架主要关注毒性、幻觉、违规等一般风险，难以充分刻画 LLM 在照护场景中回应的细微风险。RubRIX（Rubric-based Risk Index）是一种由理论驱动并经临床人员验证的评估框架，基于关怀伦理要素（Elements of an Ethic of Care），将五项实证风险维度操作化：疏忽、偏见与污名、信息不准确、无批判性认同及认知傲慢。研究使用 Reddit 和 ALZConnected 中超过 20,000 条照护者查询，评估六个前沿 LLM。采用评分标准引导改进后，各模型单轮迭代的风险组成部分均降低 45–98%。该工作为高负担场景中开发领域敏感、以用户为中心的评估框架提供了方法，并发布 benchmark 数据集，以支持 AI 辅助支持中的情境风险评估研究。

关键要点

01现有通用风险评估难以覆盖照护场景中的细微风险。
02RubRIX 将五项风险维度操作化，并由临床人员验证。
03六个前沿 LLM 在超过 20,000 条照护者查询上接受评估。
04评分标准引导的单轮改进使各模型风险组成部分降低 45–98%。
05研究发布 benchmark 数据集，支持情境风险评估。

解读

尚无解读。

原始英文摘要

arXiv:2601.13235v2 Announce Type: replace-cross Abstract: Caregivers seeking AI-mediated support express complex needs---information-seeking, emotional validation, and distress cues---that warrant careful evaluation of response safety and appropriateness. Existing AI evaluation frameworks, primarily focused on general risks (toxicity, hallucinations, policy violations, etc), may not adequately capture the nuanced risks of LLM-responses in caregiving-contexts. We introduce RubRIX (Rubric-based Risk Index), a theory-driven, clinician-validated framework for evaluating risks in LLM caregiving responses. Grounded in the Elements of an Ethic of Care, RubRIX operationalizes five empirically-derived risk dimensions: Inattention, Bias & Stigma, Information Inaccuracy, Uncritical Affirmation, and Epistemic Arrogance. We evaluate six state-of-the-art LLMs on over 20,000 caregiver queries from Reddit and ALZConnected. Rubric-guided refinement consistently reduced risk-components by 45-98% after one iteration across models. This work contributes a methodological approach for developing domain-sensitive, user-centered evaluation frameworks for high-burden contexts. Our findings highlight the importance of domain-sensitive, interactional risk evaluation for the responsible deployment of LLMs in caregiving support contexts. We release benchmark datasets to enable future research on contextual risk evaluation in AI-mediated support.