arXiv cs.AI· Yu Sun, Junhao Xu, Jiajia Shi, Zijin Yang·· 9 天前AI 评分24
类型安全不等于无错误：类型化决策模型跟随选项名称而非绑定定义
Type-Safe Is Not Error-Free: Typed Decision Models Follow the Option Name, Not the Definition Bound to It
AI 导读
研究在 Jev 和两个开放权重模型上仅改变选项名称与定义的映射，发现 yes/no 名称相比 0/1 对照组使决策翻转率最高高出 70.4 个百分点，且该差距在全部 4 种二元决策规则下均存在。yes/no 名称下重新分配还将平均 AUC 从 93.8% 降至低于随机的 23.2%，而随机字符串名称的翻转率接近中性对照组。整个过程中类型错误率保持 0%，表明类型正确的输出仍可能不遵循显式选项定义。
正文 · 原文
Abstract:Typed decision models return structured results, but output-type correctness alone does not ensure that decisions follow explicit option definitions. Each option pairs a name with a definition that defines its intended meaning; the name, however, can provide a competing semantic cue. We study this conflict in Jev and two open-weight models by changing only the name-definition mapping, leaving the question, state, and the names and definition texts themselves unchanged. We measure decision flips at the level of the selected definition, rather than the returned name. On 1200 decision tasks with task-specific definitions, decision-flip rates are up to 70.4 pp higher with yes/no names than with the 0/1 control. This gap holds across all 4 binary decision rules. With yes/no names, reassignment also lowers their mean AUC from 93.8% to a below-chance 23.2%. In the binary evaluations, random strings used as option names yield mean flip rates close to those of neutral controls across all three models, with comparable balanced accuracy before reassignment. Together, these results support option-name polarity as a contributor to decision instability beyond reassignment alone. The type-error rate remains 0% throughout, showing that type-correct outputs can still fail to follow explicit option definitions.
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2609.26758 [cs.AI] |
| (or arXiv:2609.26758v3 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2609.26758 arXiv-issued DOI via DataCite |
Submission history
From: Yu Sun [view email]
[v1]
Tue, 22 Sep 2026 17:38:12 UTC (83 KB)
[v2]
Wed, 23 Sep 2026 18:21:51 UTC (79 KB)
[v3]
Thu, 8 Oct 2026 06:34:45 UTC (79 KB)
来源：arXiv cs.AI · arxiv.org