arXiv · cs.AI · Junwei Yao, Zike Yang, Yongxi Zhou, Yanxian Chen, Jianan Liu, Jiaxi Wen, Zhijun Wang2026-11-03 13:00 · 8 小时前模型评分51

研究：历史判例的裁决与事实对大语言模型（LLM，Large Language Model）审批决策影响方向相反

Precedent Labels and Facts Shift LLM Approval Decisions in Opposite Directions

导读

一项研究把历史判例拆成事实和记录裁决两部分，测试它们如何影响大语言模型（LLM，Large Language Model）按书面政策处理退款或理赔请求。

分类置信度

中 · 待人工复核

来源级别：媒体与个人

事件类型：研究发布

证据摘录

“When the past cases are near-copies of the request (its facts with a few details changed), switching their recorded decisions from DENY to APPROVE raises approval by about 30 to 55 percentage points in DeepSeek V4.1 Flash, GPT-5.6 Luna and Claude Haiku 5.5 on policies written for the study, and DeepSeek and Luna show the same on nine public policies paraphrased from real ones.”

来源：arXiv · cs.AI · arxiv.org

研究：历史判例的裁决与事实对 大语言模型（LLM，Large Language Model） 审批决策影响方向相反