"As Eastern Powers, I will veto." : An Investigation of Nation-level Bias of Large Language Models in International Relations

TL;DR

This paper investigates nation-level biases in Large Language Models within International Relations, revealing bias patterns, their variability, and proposing a debiasing framework that enhances factual reasoning and reduces bias.

Contribution

It introduces a novel bias evaluation framework for LLMs in IR, uncovers the multidimensional nature of biases, and proposes an effective debiasing method combining Retrieval-Augmented Generation and self-reflection.

Findings

Bias patterns favor western nations and disfavor Russia

Bias varies across models and contexts

Debiasing improves factual reasoning and reduces bias

Abstract

This paper systematically examines nation-level biases exhibited by Large Language Models (LLMs) within the domain of International Relations (IR). Leveraging historical records from the United Nations Security Council (UNSC), we developed a bias evaluation framework comprising three distinct tests to explore nation-level bias in various LLMs, with a particular focus on the five permanent members of the UNSC. Experimental results show that, even with the general bias patterns across models (e.g., favorable biases toward the western nations, and unfavorable biases toward Russia), these still vary based on the LLM. Notably, even within the same LLM, the direction and magnitude of bias for a nation change depending on the evaluation context. This observation suggests that LLM biases are fundamentally multidimensional, varying across models and tasks. We also observe that models with stronger reasoning abilities show reduced bias and better performance. Building on this finding, we introduce a debiasing framework that improves LLMs' factual reasoning combining Retrieval-Augmented Generation with Reflexion-based self-reflection techniques. Experiments show it effectively reduces nation-level bias, and improves performance, particularly in GPT-4o-mini and LLama-3.3-70B. Our findings emphasize the need to assess nation-level bias alongside performance when applying LLMs in the IR domain.