Exploring the Jungle of Bias: Political Bias Attribution in Language Models via Dependency Analysis
This work addresses the issue of bias in LLMs for researchers and practitioners in AI ethics, but it is incremental as it builds on existing causal fairness frameworks without introducing a new paradigm.
The authors tackled the problem of understanding political bias in Large Language Models (LLMs) by analyzing bias through causal fairness analysis and using a prompt-based method with Activity Dependency Networks to attribute bias to confounding and mediating attributes, finding that disparate treatment in LLM ratings of political debates can be partly explained by these factors and model misalignment.
The rapid advancement of Large Language Models (LLMs) has sparked intense debate regarding the prevalence of bias in these models and its mitigation. Yet, as exemplified by both results on debiasing methods in the literature and reports of alignment-related defects from the wider community, bias remains a poorly understood topic despite its practical relevance. To enhance the understanding of the internal causes of bias, we analyse LLM bias through the lens of causal fairness analysis, which enables us to both comprehend the origins of bias and reason about its downstream consequences and mitigation. To operationalize this framework, we propose a prompt-based method for the extraction of confounding and mediating attributes which contribute to the LLM decision process. By applying Activity Dependency Networks (ADNs), we then analyse how these attributes influence an LLM's decision process. We apply our method to LLM ratings of argument quality in political debates. We find that the observed disparate treatment can at least in part be attributed to confounding and mitigating attributes and model misalignment, and discuss the consequences of our findings for human-AI alignment and bias mitigation. Our code and data are at https://github.com/david-jenny/LLM-Political-Study.