Newest papers

Humans Correct Their Judgments Through Debate, but Weaker AI Models Do Not

María Victoria Carro, Denise Alejandra Mester, Facundo Nieto, Oscar Stanchi, Guido Ernesto Bergman, Giovanni Franco Gabriel Marrafini, Trinidad Borrell, Guido Freire, Joaquin Salvador Machulsky, Mario Leiva, Ana Lopez, Nicolas Spinelli, Luca Nicolás Forziati Gangi, Eitan Sprejer, Federico Barrera Lemarchand, Gerardo I. Simari, María Vanina Martínez · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.28539 · cited by 0 works (OpenAlex) · open-access PDF

Chinese

AI安全通过辩论提出了两个AI系统争论对立的立场,一个法官,无论是人类还是一个较弱的模型,会决定哪个是正确的。通过帮助监督者提供高质量的培训信号,辩论旨在鼓励从AI系统中引导真实和安全的行为,其能力超越了我们自己的。然而,这些人类判断不是中立的:人们往往依靠先进的AI系统可以学习利用的偏见。利用从集体智能和社会认知中获得的见解,这篇论文研究了辩论是否可以引导人类和LLM监督者走向真理,尽管他们的能力超越了我们自己的信念。我们评估了几种协议配置。我们还探索了由人群的智慧所启发的设置,例如对辩论者数量和多样性的支持,以及包括辩论和加密的交互

Machine translation (Meta M2M100); the English original follows.

Abstract

How can we effectively oversee AI systems that surpass human intelligence? AI Safety via Debate proposes that two AI systems argue opposing positions and a judge, either a human or a weaker model, decides which is correct. By facilitating supervisors in providing high-quality training signals, debate aims to encourage truthful and safe behavior from AI systems whose capabilities exceed our own. However, human judgment is not neutral: people often rely on biases which advanced AI systems may learn to exploit. Drawing on insights from collective intelligence and social cognition, this paper investigates whether debate can guide human and LLM supervisors toward the truth despite their prior beliefs. We evaluate several protocol configurations. We also explore setups inspired by the wisdom of the crowd such as the number and diversity of judges, as well as interaction mechanisms including deliberation and voting. Overall, we find that debate increases accuracy relative to prior beliefs for human judges, whereas the consultancy baseline, in which a single expert argues for one answer, does not. For weaker language models, however, the results are mixed: debate provides no comparable gains in most settings and can even reduce accuracy. Among multi-judge protocols, weighted voting produces the highest accuracy for human supervisors, whereas for LLMs none of them significantly improves group accuracy, even as the number of judges increases. Notably, the behavior of minority judges in non-diverse deliberation protocols suggests that, under symmetric settings, incorrect positions could be more persuasive than correct ones. These findings support debate as a viable path toward scalable human oversight, but cast doubt on its benefits for weaker AI supervisors, raising concerns about fully automated training and control pipelines in high-stakes settings.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Deliberation, Diversity (politics), Voting, Social psychology, Psychology, Computer science

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange