通过幽默探讨语言模型中身份相关的反事实不公平性
Investigating Counterfactual Unfairness in LLMs towards Identities through Humor
AI总结：
本文通过观察在保持其他因素不变的情况下交换说话者和被指对象，研究语言模型在幽默生成拒绝、说话者意图推断及社会影响预测中的反事实不公平性，揭示身份相关的偏见模式。
摘要：
幽默揭示了社会认知：我们觉得好笑的东西往往反映我们的身份和对他人的判断。当语言模型处理幽默时，它们对幽默的反应暴露了它们从训练数据中内化的社会假设。在本文中，我们通过观察在保持其他因素不变的情况下交换说话者和被指对象，研究通过幽默探讨反事实不公平性。我们的框架涵盖三个任务：幽默生成拒绝、说话者意图推断以及关系/社会影响预测，涵盖无身份偏见的幽默和有身份偏见的冒犯性幽默。我们引入了可解释的偏见度量标准，以捕捉身份交换下的不对称模式。在最先进的模型上的实验揭示了持续的关系差异：由特权说话者讲述的笑话被拒绝的频率比非特权说话者高出多达67.5%，被判断为恶意的频率高出64.7%，并在5分制的社会伤害评分上高出1.5分。这些模式突显了生成模型中敏感性和刻板印象共存的情况，使公平性和文化契合的努力变得更加复杂。
英文摘要：
Humor holds up a mirror to social perception: what we find funny often reflects who we are and how we judge others. When language models engage with humor, their reactions expose the social assumptions they have internalized from training data. In this paper, we investigate counterfactual unfairness through humor by observing how the model's responses change when we swap who speaks and who is addressed while holding other factors constant. Our framework spans three tasks: humor generation refusal, speaker intention inference, and relational/societal impact prediction, covering both identity-agnostic humor and identity-specific disparagement humor. We introduce interpretable bias metrics that capture asymmetric patterns under identity swaps. Experiments across state-of-the-art models reveal consistent relational disparities: jokes told by privileged speakers are refused up to 67.5% more often, judged as malicious 64.7% more frequently, and rated up to 1.5 points higher in social harm on a 5-point scale. These patterns highlight how sensitivity and stereotyping coexist in generative models, complicating efforts toward fairness and cultural alignment.
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。
