记忆传染：通过智能体记忆的评估者偏见的跨时间传播
Memory Contagion: Cross-Temporal Propagation of Evaluator Bias via Agent Memory
AI总结：
研究大型语言模型智能体记忆中评估者偏见通过记忆存储跨时间传播的现象，发现即使完美记忆整合也会导致偏见传播，且无安全阈值。
摘要：
大型语言模型（LLM）智能体越来越依赖记忆系统来维持长期连贯性。最近的研究表明，智能体记忆在持续整合过程中会退化。然而，现有研究假设记忆源自无偏见的经验。在这项工作中，我们识别并形式化了一种新现象：记忆传染——评估者偏见通过智能体记忆的跨时间传播。我们表明，当智能体由有偏见的评估者训练或引导时，其经验会变得有偏见；当这些轨迹被存储并整合到记忆中时，偏见会传播给从同一记忆存储中检索的未来智能体，即使整合是完美的（神谕）。在两种偏见类型（长度偏好、权威偏见）和四个实验阶段中，我们证明了：（1）即使完美整合（神谕条件），记忆传染也会发生，证明有偏见的输入是传染的充分原因；（2）整合根据偏见类型产生相反的效果——稳健地衰减长度偏见，同时初步放大权威偏见（单次运行估计），表明存在偏见类型依赖的交互作用；（3）未观察到安全阈值：在污染率低至p=0.2时检测到偏见传播。我们的发现暴露了当前智能体记忆设计中的关键脆弱性，并提供了测量跨时间偏见传播的形式化工具。
英文摘要：
Large Language Model (LLM) agents increasingly rely on memory systems to maintain long-term coherence. Recent work shows that agent memories degrade during continuous consolidation. However, existing research assumes memories are derived from unbiased experiences. In this work, we identify and formalize a novel phenomenon: Memory Contagion -- the cross-temporal propagation of evaluator bias through agent memory. We show that when agents are trained or guided by biased evaluators, their experiences become biased; when these trajectories are stored and consolidated into memory, the bias propagates to future agents retrieving from the same memory store, even when consolidation is perfect (oracle). Across two bias types (length preference, authority bias) and four experimental phases, we demonstrate: (1) Memory Contagion occurs for length bias even with perfect consolidation on older models (Gamma_A = 13.18, DeepSeek V4-Chat), while newer models (V4-Pro, Claude) are immune, proving both that biased input is a sufficient cause and that contagion is model-generation-dependent; (2) authority bias fails to propagate in all 15 controlled multi-seed experiments (Gamma_A = 0.00), revealing that not all evaluator biases can cross temporal boundaries through current memory architectures; (3) No observed safe threshold: length bias propagation is detected at contamination rates as low as p=0.2. Our findings expose a critical but contingent vulnerability in current agent memory designs and provide formal tools for measuring cross-temporal bias propagation.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
