智能体人工智能对不一致和不完整工具响应的鲁棒性分析
Robustness Analysis of Agentic AI to Inconsistent and Incomplete Tool Responses
- Aalborg University（奥尔堡大学）
- Zhejiang University（浙江大学）
- Northeastern University（东北大学）
机构由 AI 辅助整理，请以论文原文为准。
AI总结：
该研究通过在零售客户服务领域注入受控故障，分析智能体AI对不一致、不完整工具响应的鲁棒性，发现两类响应在特定通道中可不对称识别，动作分布特征存在差异。
AI中文摘要：
对错误工具返回的鲁棒性意味着按照该返回要求的方式进行回答，这取决于工具出错的方式。发生故障的工具与返回格式正确的虚假信息的工具是不同的问题，需要不同的解决方案。我们要问的是，当返回到达时，这两者是否已经存在差异。这是一项定性试点研究：我们对单个决策点进行评分，而非让智能体运行至完成。我们在零售客户服务领域注入受控故障，并从模型的对数概率中读取两个通道：仅在工具模式下返回内容的似然度、以及在整个轨迹下的似然度，还有其在合法动作上的分布，需同时读取分布形态和概率质量所在位置。不完整的返回在所有情况下都是可识别的，在仅工具模式下，其概率处于其他任何条件都未进入的范围；且只要有移动空间，它就会将概率质量推向重新读取状态的工具。不一致的返回不会改变工具模式通道，而是在似然度比较中，在上下文已逐字携带其真实值的字段上记录，而非在整个领域策略中存在矛盾的字段上记录。动作分布为每种条件提供了独特的特征，但按返回对下一个动作的影响程度排序，而非按故障类别排序。因此识别是不对称的：每种条件在某个通道中可识别，且没有通道对所有条件都可识别。
英文摘要：
Tool-using agents increasingly rely on external tools to complete multi-step tasks, but tool returns can fail in different ways and require different recovery actions. Existing robustness studies often use uncertainty-based measures to detect when an agent becomes unreliable. These measures can reveal that something has gone wrong, but they do not directly identify the type of tool failure or the appropriate response. We address this limitation by analyzing tool failures at the moment a return enters the agent context. Our approach combines two complementary signals. The first compares the likelihood of the returned content under the tool schema and under the full trajectory prefix. The second measures the agent's probability distribution over its legal next actions. We evaluate the approach by injecting incomplete and inconsistent returns into a retail customer-service benchmark. The results show that likelihood-based signals clearly capture incomplete returns and some direct inconsistencies, while action-based signals reveal how strongly a failure changes the next decision. Some failures that are weak under likelihood signals can still redirect the agent toward state-changing actions. These findings show that tool failures can be recognized at the return boundary, but reliable diagnosis requires combining multiple signals.
