校准不等于验证：面向混合智能体的可证伪共形路由
原标题：Calibration Is Not Verification: Falsifiability-Aware Conformal Routing for Mixture-of-Agents
AI 摘要
来自LUT大学、戴尔科技和帝国理工学院的研究者提出C-MoA，一种基于智能体间语义一致性的共形过滤方法，将一致性转化为声明级非一致性分数并在样本级校准阈值，为异构Mixture-of-Agents提供无分布假设的域内事实性控制。实验显示C-MoA在长文本生成中几乎将保留声明精确率翻倍，并在无需重新校准的情况下跨域迁移，但在短文本问答中失效。作者进一步提出CONTRA-MoA，加入盲化近似错误竞赛、留一智能体稳定性和可用性感知融合，仅在验证器具备领域知识时有效，否则信号接近随机。
正文节选
Calibration Is Not Verification: Falsifiability-Aware Conformal Routing for Mixture-of-Agents Abstract Multi-agent language systems often treat agreement as evidence, yet heterogeneous agents can jointly repeat an unsupported claim or omit a correct specialist fact. We introduce C-MoA, an agreement-based conformal filter that turns inter-agent semantic support into a claim-level nonconformity score and calibrates a retention threshold at the example level, giving distribution-free within-domain