大型语言模型对意识形态操纵的易感程度如何?
How Susceptible are Large Language Models to Ideological Manipulation?
AI总结:
本文研究 LLM 是否易受意识形态操纵,发现少量意识形态化指令微调样本即可显著改变其立场,并跨主题泛化,凸显数据投毒风险与防护必要性。
摘要:
大型语言模型(LLM)有可能对公众认知及其与信息的互动产生重大影响。这引发了人们的担忧:如果这些模型中的意识形态能够被轻易操纵,可能会产生何种社会影响。在本研究中,我们调查了 LLM 从其指令微调数据中学习并泛化意识形态偏见的有效程度。研究结果揭示了一种令人担忧的脆弱性:仅接触少量受意识形态驱动的样本,就会显著改变 LLM 的意识形态。值得注意的是,LLM 表现出一种惊人的能力,能够吸收某一主题中的意识形态,并将其泛化到甚至毫不相关的主题上。LLM 的意识形态如此容易被扭曲,凸显了恶意行为者故意投毒训练数据或数据标注者无意中引入偏见所带来的风险,也强调了必须建立稳健的防护措施,以减轻意识形态操纵对 LLM 的影响。
英文摘要:
Large Language Models (LLMs) possess the potential to exert substantial influence on public perceptions and interactions with information. This raises concerns about the societal impact that could arise if the ideologies within these models can be easily manipulated. In this work, we investigate how effectively LLMs can learn and generalize ideological biases from their instruction-tuning data. Our findings reveal a concerning vulnerability: exposure to only a small amount of ideologically driven samples significantly alters the ideology of LLMs. Notably, LLMs demonstrate a startling ability to absorb ideology from one topic and generalize it to even unrelated ones. The ease with which LLMs' ideologies can be skewed underscores the risks associated with intentionally poisoned training data by malicious actors or inadvertently introduced biases by data annotators. It also emphasizes the imperative for robust safeguards to mitigate the influence of ideological manipulations on LLMs.
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。