大语言模型中的政治倾向：心理测量身份与行为偏见的多维审计
Political Alignment in Large Language Models: A Multidimensional Audit of Psychometric Identity and Behavioral Bias
AI总结：
研究通过多维心理测量工具审计26个大语言模型，发现模型在政治倾向上聚类于自由主义左 quadrant，且模型身份解释了大部分变异性，但心理测量意识形态未显著预测分类误差。
摘要：
随着大语言模型（LLMs）的广泛应用，理解其政治立场表达对于评估对齐性和下游影响至关重要。我们使用三种政治心理测量量表（政治罗盘、SapplyValues、8Values）和新闻偏见标注任务审计了26个现代LLMs。为测试鲁棒性，量表在多个语义提示变体下进行测试，并通过两因素方差分析分离模型和提示效应。大多数模型聚集在相似的意识形态区域，96.3%位于政治罗盘的自由主义左象限，且模型身份解释了大部分提示变异性（η²>0.90）。跨仪器比较表明，政治罗盘的社会轴比与权威相关的测量更强烈地与文化进步主义相关（r=-0.64）。我们观察到开源与闭源模型之间的差异以及在下游分类中检测极端政治偏见的不对称表现。回归分析发现，心理测量意识形态定位不显著预测分类误差，提供没有统计学显著关系的证据，表明对话意识形态身份与任务层面行为之间没有显著关系。这些发现表明，单轴评估是不够的，多维审计框架对于表征部署LLMs中的对齐行为至关重要。我们的代码和数据可在https://github.com/sakhadib/PolAlignLLM公开获取。
英文摘要：
As large language models (LLMs) are increasingly deployed, understanding how they express political positioning is important for evaluating alignment and downstream effects. We audit 26 contemporary LLMs using three political psychometric inventories (Political Compass, SapplyValues, 8Values) and a news bias labeling task. To test robustness, inventories are administered across multiple semantic prompt variants and analyzed with a two-way ANOVA separating model and prompt effects. Most models cluster in a similar ideological region, with 96.3% located in the Libertarian-Left quadrant of the Political Compass, and model identity explaining most variance across prompt variants ($η^2 > 0.90$). Cross-instrument comparisons suggest that the Political Compass social axis aligns more strongly with cultural progressivism than authority-related measures ($r=-0.64$). We observe differences between open-weight and closed-source models and asymmetric performance in detecting extreme political bias in downstream classification. Regression analysis finds that psychometric ideological positioning does not significantly predict classification errors, providing no evidence of a statistically significant relationship between conversational ideological identity and task-level behavior. These findings suggest that single-axis evaluations are insufficient and that multidimensional auditing frameworks are important to characterize alignment behavior in deployed LLMs. Our code and data are publicly available at https://github.com/sakhadib/PolAlignLLM.
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。
