缓解生成式AI自适应系统中的不确定性交互：愿景、挑战与初步指南
原标题：Mitigating Uncertainty Interactions in GenAI-based Adaptive Systems: Vision, Challenges and Preliminary Guidelines
AI 摘要
这篇 arXiv 立场论文提出一个概念框架，用于刻画和缓解基于生成式 AI（GAI）的软件密集型系统中的不确定性交互。作者指出 LLM 与 agentic 子系统带来概率输出、幻觉、上下文窗口限制和记忆陈旧等新型不确定性，这些不确定性会跨组件交互和传播。论文引入「缓解函数」抽象，将不确定性交互类别、系统设计特征、关键性与架构策略映射到质量属性及权衡，并在 LLM 服务、AI 驱动运维和 agentic 云管理三个用例中做概念验证。
正文节选
Mitigating Uncertainty Interactions in GenAI-based Adaptive Systems: Vision, Challenges and Preliminary Guidelines Abstract Modern software-intensive systems increasingly incorporate GAI (GAI) components, including LLM and agentic subsystems, which introduce novel and compounding sources of uncertainty across system layers. The self-adaptive systems community has made significant strides in understanding and managing uncertainty. The intrinsic characteristics of GAI, including probabilistic outp