并排比较加剧语言模型中的方言偏见
Side-by-side Comparison Amplifies Dialect Bias in Language Models
AI总结：
本研究通过并排比较标准美式英语和非裔美国英语的推文，发现语言模型中的隐性方言偏见在对比设置下显著加剧，且显性方言偏见在安全对齐微调后仍存在。
摘要：
语言模型（LMs）可能因其方言变体而表现出偏见，即使在没有方言标签的情况下，这种行为被称为隐性方言偏见。在这项工作中，我们通过评估语言模型如何将刻板特征（源自社会心理学关于种族偏见的研究）与标准美式英语（SAE）和非裔美国英语（AAVE）中意图等效的推文相关联，来量化在线话语中的隐性方言偏见。虽然先前的研究表明，在单独评估推文时，语言模型将更多负面刻板印象与AAVE关联，但我们惊讶地发现，当SAE/AAVE推文对并排比较时，这种偏见显著加剧，这种设置更接近模型用于排名候选人的高影响力决策环境。当明确指定方言标签时，偏见只会恶化。考虑到商业开发者为了减轻其语言模型中的偏见所做的广泛努力，这一点令人震惊。令人鼓舞的是，我们表明反事实公平微调可以减轻某些刻板特征的隐性方言偏见，减少单独评估推文时的平均差异，然而，在并排评估SAE/AAVE推文时，这些改进并不一致地适用于所有特征。我们的发现表明，现有的隐性方言偏见评估设置可能低估了其严重性，特别是在对比设置中。此外，即使在安全对齐微调后，显性方言偏见仍然显著，表明它仍然是一个未解决的问题，并激励需要更稳健的评估和缓解框架。
英文摘要：
Language models (LMs) can exhibit biases based on variations in their dialects, even in the absence of a dialect label, a behavior known as covert dialect bias. In this work, we quantify covert dialect bias in online discourse by evaluating how LMs associate stereotypical traits (derived from social psychology research on racial bias) with intent-equivalent tweets in Standard American English (SAE) and African-American Vernacular English (AAVE). While prior work shows that LMs associate more negative stereotypes with AAVE when evaluating tweets in isolation, we are surprised to find that this bias is significantly exacerbated when SAE / AAVE tweet pairs are compared side by side, a setting that more closely reflects high-impact decision making contexts in which models are used to rank candidates. The bias only worsens when dialect labels are explicitly specified. This is striking, given the extensive efforts from commercial developers to mitigate bias in their LMs. Encouragingly, we show that counterfactual fairness finetuning can mitigate covert dialect bias for some stereotypical traits, reducing average disparities when evaluating tweets in isolation, however, these improvements do not consistently hold across traits when evaluating SAE / AAVE tweets side by side. Our findings show that existing evaluation settings for covert dialect bias may underestimate its severity, specifically in contrastive settings. Additionally, overt dialect bias remains pronounced even after safety aligned finetuning, indicating that it remains an unresolved problem, and motivates the need for more robust evaluation and mitigation frameworks.
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。
