CIDER: A Causal Cure for Brand-Obsessed Text-to-Image Models
This addresses ethical and legal risks for users and developers of generative AI by providing a practical, inference-time solution to mitigate brand bias without costly retraining.
The paper tackles the problem of 'brand bias' in text-to-image models, where models generate content featuring commercial brands from generic prompts, and proposes CIDER, a model-agnostic framework that reduces this bias by 30-50% on leading models while maintaining image quality.
Text-to-image (T2I) models exhibit a significant yet under-explored "brand bias", a tendency to generate contents featuring dominant commercial brands from generic prompts, posing ethical and legal risks. We propose CIDER, a novel, model-agnostic framework to mitigate bias at inference-time through prompt refinement to avoid costly retraining. CIDER uses a lightweight detector to identify branded content and a Vision-Language Model (VLM) to generate stylistically divergent alternatives. We introduce the Brand Neutrality Score (BNS) to quantify this issue and perform extensive experiments on leading T2I models. Results show CIDER significantly reduces both explicit and implicit biases while maintaining image quality and aesthetic appeal. Our work offers a practical solution for more original and equitable content, contributing to the development of trustworthy generative AI.