Large Language Models (LLMs) exhibit social biases, which can lead to harmful stereotypes and unfair outcomes. We propose \textbf{Multi-Persona Thinking (MPT)}, a simple inference-time framework that reduces social bias by encouraging reasoning from multiple perspectives. MPT guides the model to consider contrasting social identities, such as male and female, together with a neutral viewpoint. These viewpoints then interact through an iterative reasoning process to identify and correct biased judgments. This design transforms the potential weakness of persona assignment into a mechanism to mitigate bias. We evaluate MPT on two widely used bias benchmarks with both open-source and closed-source models. Our results show that MPT achieves a lower bias than the existing prompting-based methods while maintaining the core reasoning ability.
@article{arxiv.2601.15488,
title = {Multi-Persona Thinking for Bias Mitigation in Large Language Models},
author = {Yuxing Chen and Guoqing Luo and Zijun Wu and Lili Mou},
journal= {arXiv preprint arXiv:2601.15488},
year = {2026}
}