这篇论文用GPT-4o和DeepSeek-chat做实验，发现AI自我评估时策略偏好会崩坏，还跨模态传染，数字很硬核，搞多模态和智能体的值得看。
论文发现，当AI智能体使用语言模型在反馈循环中自我评估时，会产生系统性偏好偏差。在多模态环境中，评估者偏好崩溃（EPC）被显著放大：用GPT-4o评估DeepSeek-chat时，单一策略step_by_step吸收48.4%权重，是纯文本自评估崩溃程度的3.2倍，而三个视觉域策略合计仅占9.1%。跨模态传染现象揭示：在一个模态获得的评估者偏好会转移到另一个模态并扭曲策略选择。四阶段隔离训练实验测量了传染系数，并发现跨模态暴露后最优策略发生反转。在53次独立重复、15,592次API调用统计中，跨模型评估产生强对称双向传染（均值γ_{T->V}=1.176, γ_{V->T}=1.089），而自评估（DeepSeek-chat）97%运行传染为零。
Multimodal Evaluator Preference Collapse: Cross-Modal Contagion in Self-Evolving Agents
When AI agents use language models to evaluate their own outputs in a feedback loop, systematic biases emerge. We show that Evaluator Preference Collapse (EPC) is dramatically amplified in multimodal settings. Using GPT-4o to evaluate DeepSeek-chat across text and visual tasks, we find that a single strategy (step_by_step) absorbs 48.4% of all weight -- 3.2x the collapse observed in text-only self-evaluation -- while three visual-domain strategies receive only 9.1% combined weight. We then demonstrate a novel phenomenon we term cross-modal contagion: evaluator preferences acquired on one modality transfer to and corrupt strategy selection on another. Through a four-phase isolation training paradigm, we measure contagion coefficients and document strategy inversion -- the optimal strategy for a modality reverses after cross-modal exposure. A Phase 3 statistical validation across four evaluator configurations (N=53 total independent repetitions, 15,592 API calls) reveals a clear hierarchy: cross-model evaluation (GPT-4o, N=8) produces strong but symmetric bidirectional contagion (mean gamma_{T->V}=1.176, gamma_{V->T}=1.089, Delta=-0.088, p=0.575, Cohen's d=0.29); high round counts (DashScope, 50 rounds) cause collapse to single-strategy dominance (70% zero contagion); and self-evaluation provides near-complete immunity -- 97% of runs (N=30, DeepSeek-chat) yield exactly zero contagion (mean gamma=0.033, 95% CI [-0.031, 0.010], p=0.642, d=0.07). No evaluator condition shows statistically significant directional asymmetry. We introduce the contagion matrix indexed by evaluator identity, release the MM-EPC experimental framework, and identify cross-model evaluator architecture as the primary risk factor for preference contagion.