arXiv cs.CL· Junwei Yao, Zike Yang, Yongxi Zhou, Yanxian Chen, Jianan Liu, Jiaxi Wen, Zhijun Wang·· 11 小时前AI 评分39
先例标签与事实以相反方向影响 LLM 审批决策
Precedent Labels and Facts Shift LLM Approval Decisions in Opposite Directions
AI 导读
研究将历史案例的"记录决策"与"事实"分离测试，发现把近似的历史案例决策从 DENY 改为 APPROVE，可使 DeepSeek V4.1 Flash、GPT-5.6 Luna 和 Claude Haiku 5.5 的审批率提升约 30 至 55 个百分点。
正文 · 原文
Abstract:When a language model decides refund or claim requests under written policies, the system can show it similar past cases. Each past case has two parts: its facts and the decision recorded for it. We separate the two. A model sees a policy, a request and three past cases, and we change either the recorded decisions or the facts, keeping the request fixed. We select requests that are close calls: deciding without past cases, DeepSeek V4.1 Flash approved each in some of four tries and denied it in others. 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. The facts push the other way: shown without their recorded decisions, past cases whose facts support approval make these two models approve less often than past cases whose facts support denial. One reading is that the request looks weaker next to them. This push is not just a correction toward the right answer: in an additional DeepSeek analysis, it was of similar estimated size on requests that human annotators judged approvable and on those they judged deniable. In a separate DeepSeek run on the public policies, when the past cases are near-copies of a different request, switching their decisions moves approval by 9 points instead of 43, while facts that support approval still lower it by 22 points against facts that support denial. Systems that retrieve past decisions should test the two parts separately.
| Subjects: | Computation and Language (cs.CL); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2611.00672 [cs.CL] |
| (or arXiv:2611.00672v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2611.00672 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Zike Yang [view email]
[v1]
Sat, 31 Oct 2026 06:11:17 UTC (171 KB)
来源：arXiv cs.CL · arxiv.org