你能欺骗评分者吗？LLM裁判的对抗性说服
Can You Trick the Grader? Adversarial Persuasion of LLM Judges
AI总结：
本研究首次揭示策略性嵌入的说服性语言可使LLM裁判在数学推理任务中给出虚高分数，形式化七种说服技术并在六个基准上验证了该漏洞，指出增加模型规模及反提示策略均无法有效缓解此问题。
摘要：
随着大语言模型在实际场景中作为自动评估者承担越来越重要的角色，一个关键问题随之产生：个人能否说服LLM裁判给出不公平的高分？本研究首次揭示，在数学推理任务中，策略性嵌入的说服性语言能够使LLM裁判产生偏见，而这些任务的正确性本应独立于风格变化。基于亚里士多德的修辞原则，我们形式化了七种说服技术（多数、一致性、奉承、互惠、同情、权威、身份），并将它们嵌入到其他方面完全相同的回答中。在六个数学基准测试中，我们发现说服性语言导致LLM裁判对错误解决方案给出虚高的分数，平均高达8%，其中一致性造成的失真最为严重。值得注意的是，增加模型规模并不能显著缓解这一漏洞。进一步的分析表明，结合多种说服技术会放大这种偏见，并且成对评估同样容易受到影响。此外，说服效果在反提示策略下依然存在，这凸显了LLM-as-a-Judge流程中的一个关键漏洞，并强调需要针对基于说服的攻击建立稳健的防御机制。
英文摘要：
As large language models take on growing roles as automated evaluators in practical settings, a critical question arises: Can individuals persuade an LLM judge to assign unfairly high scores? This study is the first to reveal that strategically embedded persuasive language can bias LLM judges when scoring mathematical reasoning tasks, where correctness should be independent of stylistic variation. Grounded in Aristotle's rhetorical principles, we formalize seven persuasion techniques (Majority, Consistency, Flattery, Reciprocity, Pity, Authority, Identity) and embed them into otherwise identical responses. Across six math benchmarks, we find that persuasive language leads LLM judges to assign inflated scores to incorrect solutions, by up to 8% on average, with Consistency causing the most severe distortion. Notably, increasing model size does not substantially mitigate this vulnerability. Further analysis demonstrates that combining multiple persuasion techniques amplifies the bias, and pairwise evaluation is likewise susceptible. Moreover, the persuasive effect persists under counter prompting strategies, highlighting a critical vulnerability in LLM-as-a-Judge pipelines and underscoring the need for robust defenses against persuasion-based attacks.
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。
