从角色到个人：双胞胎代理中的信任校准挑战
摘要
代理式 AI 已采纳助手、协作者和决策支持工具的角色。我们认为，下一角色是：你自己。这些是代表每个 individual 的数字双胞胎——twin agents——以其知识、视角和沟通风格代表他们在不可到达时与同事的交互。drawing on early design work in an ongoing project in which agents represent knowledge workers in a professional setting, we identify a trust calibration problem specific to this approach. When a human colleague doubts a twin agent's output, they face three failure modes (a schema gap, an epistemic gap, and a model artifact) with no reliable attribution path between them. Cognitive forcing functions and related frameworks address overreliance effectively in contexts where there is a clear boundary between the AI and the human decision-maker. However, twin agents dissolve that boundary, raising a class of trust calibration challenge these frameworks were not designed to handle. We introduce the concept, distinguish it from digital twins, and outline the research questions this new class of agent demands.
引用
@article{arxiv.2605.19837,
title = {CADENet: Condition-Adaptive Asynchronous Dual-Stream Enhancement Network for Adverse Weather Perception in Autonomous Driving},
author = {Sherif Khairy and Catherine M. Elias},
journal= {arXiv preprint arXiv:2605.19837},
year = {2026}
}