沉默的偏见：LLMs所学习到的拒绝的黑暗面
Silenced Biases: The Dark Side LLMs Learned to Refuse
AI总结：
本文提出Silenced Bias Benchmark，通过激活引导减少问答中的拒绝，揭示模型潜在的公平性问题，挑战传统公平评估方法。
摘要：
安全对齐的大语言模型（LLMs）正变得越来越普遍，尤其是在敏感应用中，公平性至关重要，偏见输出可能造成重大损害。然而，评估模型公平性是一个复杂挑战，现有方法通常使用标准问答（QA）风格方案。此类方法往往忽视深层问题，将模型的拒绝响应解释为积极的公平性测量，造成虚假公平感。本文引入沉默偏见的概念，即模型潜在空间中编码的不公平偏好，通过安全对齐有效隐藏。以往考虑类似间接偏见的方法通常依赖提示操纵或手工隐式查询，存在扩展性有限和污染评估过程的风险。我们提出Silenced Bias Benchmark（SBB），通过激活引导减少问答中的拒绝，以揭示这些偏见。SBB支持轻松扩展到新的人口群体和主题，提供一个公平性评估框架，促进未来公平模型和工具的发展，超越对齐训练的遮蔽效应。我们在多个LLMs上展示我们的方法，发现模型直接响应与潜在公平性问题之间存在令人担忧的差异。
英文摘要：
Safety-aligned large language models (LLMs) are becoming increasingly widespread, especially in sensitive applications where fairness is essential and biased outputs can cause significant harm. However, evaluating the fairness of models is a complex challenge, and approaches that do so typically utilize standard question-answer (QA) styled schemes. Such methods often overlook deeper issues by interpreting the model's refusal responses as positive fairness measurements, which creates a false sense of fairness. In this work, we introduce the concept of silenced biases, which are unfair preferences encoded within models' latent space and are effectively concealed by safety-alignment. Previous approaches that considered similar indirect biases often relied on prompt manipulation or handcrafted implicit queries, which present limited scalability and risk contaminating the evaluation process with additional biases. We propose the Silenced Bias Benchmark (SBB), which aims to uncover these biases by employing activation steering to reduce model refusals during QA. SBB supports easy expansion to new demographic groups and subjects, presenting a fairness evaluation framework that encourages the future development of fair models and tools beyond the masking effects of alignment training. We demonstrate our approach over multiple LLMs, where our findings expose an alarming distinction between models' direct responses and their underlying fairness issues.
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。
