paper-with-me

홈 › Papers

VIGNETTE: Socially Grounded Bias Evaluation for Vision-Language Models

2025-05-28 · Chahat Raj, Bowen Wei, Aylin Caliskan, Antonios Anastasopoulos, Ziwei Zhu

While bias in large language models (LLMs) is well-studied, similar concerns in vision-language models (VLMs) have received comparatively less attention. Existing VLM bias studies often focus on portrait-style images and gender-occupation associations, overlooking broader and more complex social stereotypes and their implied harm. This work introduces VIGNETTE, a large-scale VQA benchmark with 30M+ images for evaluating bias in VLMs through a question-answering framework spanning four directions: factuality, perception, stereotyping, and decision making. Beyond narrowly-centered studies, we assess how VLMs interpret identities in contextualized settings, revealing how models make trait and capability assumptions and exhibit patterns of discrimination. Drawing from social psychology, we examine how VLMs connect visual identity cues to trait and role-based inferences, encoding social hierarchies, through biased selections. Our findings uncover subtle, multifaceted, and surprising stereotypical patterns, offering insights into how VLMs construct social meaning from inputs.

📄 PDF Abstract BibTeX arXiv:2505.22897

Code (1)

chahatraj/Vignette 공식 구현 pytorch

Tasks

Decision MakingQuestion AnsweringVisual Question Answering (VQA)

Methods 이 논문이 사용한 방법론

Focus 설명 없음

Similar Papers 제목 키워드 기반

The Need for a Socially-Grounded Persona Framework for User Simulation

2026-01-12 · Pranav Narayanan Venkit, Yu Li, Yada Pruksachatkun, Chien-Sheng Wu arxiv

Synthetic personas are widely used to condition large language models (LLMs) for social simulation, yet most personas are still constructed from coarse sociodemographic attributes or summaries. We revisit persona creatio…

Enabling Scalable Evaluation of Bias Patterns in Medical LLMs

2024-10-18 · Hamed Fayyaz, Raphael Poulain, Rahmatollah Beheshti

Large language models (LLMs) have shown impressive potential in helping with numerous medical challenges. Deploying LLMs in high-stakes applications such as medicine, however, brings in many concerns. One major area of c…

Knowledge GraphsSpecificity

You Never Know: Quantization Induces Inconsistent Biases in Vision-Language Foundation Models

2024-10-26 · Eric Slyman, Anirudh Kanneganti, Sanghyun Hong, Stefan Lee

We study the impact of a standard practice in compressing foundation vision-language models - quantization - on the models' ability to produce socially-fair outputs. In contrast to prior findings with unimodal models tha…

Quantization

Serendipity with Generative AI: Repurposing knowledge components during polycrisis with a Viable Systems Model approach

2025-12-01 · Gordon Fletcher, Saomai Vu Khan arxiv

Organisations face polycrisis uncertainty yet overlook embedded knowledge. We show how generative AI can operate as a serendipity engine and knowledge transducer to discover, classify and mobilise reusable components (mo…

Misaligned by Reward: Socially Undesirable Preferences in LLMs

2026-05-06 · Gayane Ghazaryan, Esra Dönmez arxiv

Reward models are a key component of large language model alignment, serving as proxies for human preferences during training. However, existing evaluations focus primarily on broad instruction-following benchmarks, prov…