AIJul 30, 2025

FairReason: Balancing Reasoning and Social Bias in MLLMs

arXiv:2507.23067v23 citationsh-index: 10
Originality Incremental advance
AI Analysis

This addresses the trade-off between fairness and capability in MLLMs, offering practical guidance for developers, though it is incremental as it builds on existing bias-mitigation strategies.

The study tackled the problem of balancing reasoning ability and social bias mitigation in Multimodal Large Language Models, finding that a 1:4 mix of debias-focused and reasoning-centric samples trained with reinforcement learning reduced stereotype scores by 10% while retaining 88% of original reasoning accuracy.

Multimodal Large Language Models (MLLMs) already achieve state-of-the-art results across a wide range of tasks and modalities. To push their reasoning ability further, recent studies explore advanced prompting schemes and post-training fine-tuning. Although these techniques improve logical accuracy, they frequently leave the models' outputs burdened with pronounced social biases. Clarifying how reasoning gains interact with bias mitigation-and whether the two objectives inherently trade off-therefore remains an open and pressing research problem. Our study begins by benchmarking three bias-mitigation strategies-supervised fine-uning (SFT), knowledge distillation (KD), and rule-based reinforcement learning (RL)-under identical conditions, establishing their baseline strengths and weaknesses. Building on these results, we vary the proportion of debias-focused and reasoning-centric samples within each paradigm to chart the reasoning-versus-bias trade-off. Our sweeps reveal a consistent sweet spot: a roughly 1:4 mix trained with reinforcement learning cuts stereotype scores by 10% while retaining 88% of the model's original reasoning accuracy, offering concrete guidance for balancing fairness and capability in MLLMs.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes