LLM评判器在真实对话中漏检超80%缺陷，因评分标准缺失状态追踪维度。
Catching One in Five: LLM-as-Judge Blind Spots in Production Multi-Turn Transaction Agents
- 用人工标注为基准，检验部署级对话代理的LLM评判器表现
- 实际只捕获22%系统性缺陷，零漏检率批次含23个真实问题
- 关键短板是评分维度缺状态跟踪、容错等行为维度，非感知能力不足
LLM作为对话代理评估工具虽被广泛使用，但其可靠性常以与人类评分的一致性衡量，而非真实缺陷的召回率。我们研究了一个已上线的多轮餐饮订单代理，通过人工全量转录审校作为基准，测量其内置的LLM评判器实际能发现多少真实质量问题。在三个批次中，该评判器仅捕获不到四分之一的人工确认系统性问题——一个批次中9个模式仅发现2个（22%），另一批次中100轮人工确认23个缺陷及7种新交叉模式，但评判器操作门禁标记为零。分析显示失败具有结构性：评判器可识别单轮局部问题（如虚构数据、语言错误），但遗漏跨轮状态问题（如确认门锁死、购物车幻觉、升级锁死、过期指代）。根本原因在于评分标准仅包含三个粗粒度维度（意图、品牌语音、个性化），缺乏对状态追踪、安全边界、恢复机制等行为维度的分类。故障本质是路由机制而非感知能力：114个原始备注描述确认门或购物车状态问题的案例中，113个被误判为“品牌语音”，无一触发操作失败；原因是门禁逻辑依赖挂起和强断言，而非评分规则。因此当报告缺陷率为零时，罗根-格拉登修正失效（无法恢复真实率）；而当有报告缺陷时，同一估算方法表明存在3-6倍低估。对于生产级多轮代理，自动化评判仅为最低底线，不能替代人工审查。
原文摘要 · Abstract (English)
LLM-as-judge is the default instrument for evaluating conversational agents, yet its reliability is almost always reported as agreement with human ratings, not recall of real defects. We study a deployed multi-turn food-and-beverage ordering agent and measure how many genuine quality problems its built-in LLM judge catches, using exhaustive human transcript review as ground truth. Across three batches the judge surfaces well under a quarter of human-confirmed systematic problems -- 2 of 9 patterns (22%) in one batch, and its operational gate flagged zero of 100 rounds in a batch where humans confirmed 23 distinct defects and 7 new cross-cutting patterns. Our blind-spot taxonomy shows the failure is structured, not random: the judge catches turn-local issues (a fabricated statistic, a wrong language) but misses cross-turn state issues (confirm-gate lockout, cart hallucination, escalation lockout, stale referents). The mechanism: the scoring rubric exposes only three coarse axes (intent, brand-voice, personalization) and has no category for the behavioural dimensions -- state-tracking, guardrails, recovery -- where most defects cluster. The failure is routing, not perception: 113 of 114 rounds whose raw judge note describes a confirm-gate or cart-state defect are scored "brand voice", and none reach an operational failure -- the gate is wired to hangs and hard assertions, not the rubric -- so the 0% is a routing-and-wiring failure, not blindness. The consequence for prevalence estimation is sharp: when the apparent defect rate is zero the Rogan-Gladen correction degenerates -- no signal can recover the true rate -- while where the gate reports a nonzero rate the same estimator implies a 3-6x undercount under our measured sensitivity. For production multi-turn agents, automated judging is a regression floor, not a substitute for human review.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。