进步的幻象：重新评估LLM中的幻觉检测
The Illusion of Progress: Re-evaluating Hallucination Detection in LLMs
AI总结：
本文指出依赖ROUGE评估LLM幻觉检测方法存在严重误导，使用LLM-as-Judge等人类对齐指标发现现有方法性能显著下降，且简单启发式方法可媲美复杂技术，呼吁采用语义感知的稳健评估框架。
摘要：
大型语言模型（LLMs）彻底改变了自然语言处理，但其产生幻觉的倾向对可靠部署构成了严峻挑战。尽管存在多种幻觉检测方法，但其评估通常依赖于ROUGE，这是一种基于词汇重叠的指标，与人类判断不一致。通过全面的人类研究，我们证明，尽管ROUGE表现出高召回率，但其极低的精确率会导致误导性的性能估计。事实上，当使用如LLM-as-Judge等与人类对齐的指标进行评估时，几种成熟的检测方法的性能下降了高达45.9%。此外，我们的分析表明，基于响应长度的简单启发式方法可以与复杂的检测技术相媲美，暴露了当前评估实践的根本缺陷。我们认为，采用语义感知和稳健的评估框架对于准确衡量幻觉检测方法的真实性能至关重要，最终确保LLM输出的可信度。
英文摘要：
Large language models (LLMs) have revolutionized natural language processing, yet their tendency to hallucinate poses serious challenges for reliable deployment. Despite numerous hallucination detection methods, their evaluations often rely on ROUGE, a metric based on lexical overlap that misaligns with human judgments. Through comprehensive human studies, we demonstrate that while ROUGE exhibits high recall, its extremely low precision leads to misleading performance estimates. In fact, several established detection methods show performance drops of up to 45.9\% when assessed using human-aligned metrics like LLM-as-Judge. Moreover, our analysis reveals that simple heuristics based on response length can rival complex detection techniques, exposing a fundamental flaw in current evaluation practices. We argue that adopting semantically aware and robust evaluation frameworks is essential to accurately gauge the true performance of hallucination detection methods, ultimately ensuring the trustworthiness of LLM outputs.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
