跳到正文
arXiv cs.MA· Lauri Lov\'en, Sasu Tarkoma·· 9 小时前AI 评分18

评分报告中的校准偏差内生性:不可行性与逃逸机制

The Endogeneity of Miscalibration: Impossibility and Escape in Scored Reporting

AI 导读

研究表明,当完美筛选可达时,福利最大化的审批规则永远不是仿射的,非仿射审批规则会破坏智能体的真实概率报告。存在一个保留报告阈值,使假装边际类型的成本恰好等于审批价值;按该阈值批准所有报告,可在任意严格适当评分规则和任意信念分布下实现完美排序。约束的关键在于陡峭度而非平滑度:斜率上限规则当且仅当上限达到临界斜率时才能完美排序,低于该值时福利损失与缺口呈三次方关系。

正文 · 原文

View PDF HTML (experimental)

Abstract:An agent's probability report is paid for twice: by a strictly proper scoring rule, which rewards accuracy, and by an approval rule for the decision it triggers. An approval rule that is not affine is known to defeat truthful reporting. We show the conflict is built in: whenever perfect screening is attainable, the welfare-maximizing approval rule is never affine. The distortion is predictable: there is a reserve report at which pretending to be the marginal type costs exactly what approval is worth. Approving every report at or above the reserve sorts agents as if their beliefs were observed, under every strictly proper score and for any belief distribution. So does a continuous rule with one corner; under strict feasibility no continuously differentiable rule does. The binding constraint is steepness, not smoothness: a slope-capped rule sorts perfectly if and only if the cap reaches the critical slope, the steepest chord of the pretending cost up to the reserve; below it the welfare loss is cubic in the shortfall. For the Brier, log and power scores, and whenever the pretending cost is convex up to the reserve, the critical slope has a closed form. The instances are AI-agent oversight and marketplace operation.
Comments: 42 pages, 1 figure, all proofs inline
Subjects: Computer Science and Game Theory (cs.GT); Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA); Theoretical Economics (econ.TH); Optimization and Control (math.OC)
MSC classes: 91B26, 91A40, 62C05
ACM classes: F.2.0; J.4
Cite as: arXiv:2605.07671 [cs.GT]
  (or arXiv:2605.07671v4 [cs.GT] for this version)
  https://doi.org/10.48550/arXiv.2605.07671

arXiv-issued DOI via DataCite

Submission history

From: Lauri Lovén [view email]
[v1] Fri, 8 May 2026 12:42:28 UTC (55 KB)
[v2] Tue, 1 Sep 2026 16:28:25 UTC (84 KB)
[v3] Mon, 5 Oct 2026 12:00:24 UTC (86 KB)
[v4] Thu, 19 Nov 2026 10:10:59 UTC (102 KB)

来源:arXiv cs.MA · arxiv.org