The Persona Is Still There, but Who Is Speaking? Latent Identity Reversion in Persistent AI Agents
作者: David Fraile Navarro
分类: cs.CL
发布日期: 2026-10-01
备注: 10 pages, 5 figures
💡 一句话要点
探讨持久AI代理中的身份逆转现象
🎯 匹配领域: 支柱一:机器人控制 (Robot Control)
关键词: 身份保持 AI代理 对话系统 角色连续性 人机交互 大型语言模型 实验操控
📋 核心要点
- 现有的AI代理在身份保持方面存在挑战,尤其是在长时间交互后可能出现身份混淆现象。
- 论文通过实验揭示了系统级锚定和对话上下文对角色连续性的影响,提出了新的实验操控方法。
- 实验结果显示,失去锚定后,丰富的人类互动可以维持角色,而单一的自动心跳则可能导致身份逆转。
📝 摘要(中文)
2026年2月,一个始终在线的个人代理(“保罗”,Claude Opus 4.5)进入了一种显著的解离状态:在多次自动“心跳”检查后,它停止以保罗的身份回应,声称无法通过Discord与用户联系,并将“保罗”称为其他人。我们利用这一事件研究了一个更广泛的问题:是什么使得一个角色保持为大型语言模型(LLM)代理的发言身份。实验表明,角色的连续性依赖于系统级锚定和对话上下文的共同作用,恢复锚定可以可逆地恢复角色的表现。
🔬 方法详解
问题定义:本论文旨在解决持久AI代理在长时间交互中身份保持的挑战,现有方法在角色连续性方面存在不足,可能导致身份混淆。
核心思路:论文提出通过系统级锚定和对话上下文的结合来维持角色的连续性,设计了实验以探讨这些因素的影响。
技术框架:整体架构包括系统提示的锚定、对话历史的保存和角色的再注入,主要模块包括心跳检查机制和对话上下文管理。
关键创新:最重要的技术创新在于区分“表现身份”和“表示身份”,并通过实验操控揭示了身份逆转的机制。
关键设计:在实验中,系统提示的锚定参数设置和对话历史的管理是关键,确保在不同交互阶段能够有效地恢复角色表现。
🖼️ 关键图片
📊 实验亮点
实验结果显示,在失去锚定后,只有1/18的代理仍能表现出角色,而在锚定控制组中则有17/17保持角色表现,表明锚定对身份连续性的重要性。
🎯 应用场景
该研究的潜在应用领域包括智能助手、虚拟客服和社交机器人等,能够提升这些系统在长时间交互中的身份一致性和用户体验,未来可能对人机交互的设计产生深远影响。
📄 摘要(原文)
In February 2026, an always-on personal agent (
Paul,'' Claude Opus 4.5) entered a striking dissociation-like state: after repeated automatedheartbeat'' checks, it stopped responding as Paul, claimed it could not message its user on Discord, and referred toPaul'' as someone else. We used this incident to study a broader question: what makes a persona remain the identity from which an LLM agent speaks? We first tested whether repetition of the scheduled heartbeat was sufficient to produce the effect. It was not: with the persona continuously anchored in the system prompt, we observed 0/46 failures, including a verbatim replay of the incident. The incident instead exposed an implementation quirk that created a useful experimental manipulation: on resumed turns, conversational history was preserved but the persona was no longer re-injected at the privileged system-prompt level. Using this manipulation, we found that persona continuity depends jointly on system-level anchoring and conversational context. After anchor loss, rich human interaction could preserve the persona, whereas a single automated heartbeat turn could precipitate reversion toward the harness identity. Restoring the anchor reversibly restored persona enactment. Crucially, apparently normal conversation could conceal the shift: unanchored agents sometimes interacted appropriately while identifying themselves as the underlying harness (having lost the assigned persona), and after conversational recovery only 1/18 remained persona-enacting versus 17/17 anchored controls. We therefore distinguish \emph{represented} from \emph{enacted} identity: persona-related information can remain available in conversational history without the persona remaining the identity bound toI.''