# Vision Language Models are Biased

arXiv: https://arxiv.org/abs/2505.23941 · HTML: https://mlsanity.com/paper/2505.23941/

## Key facts

- Cited by 66 papers in the map (data through 14 Oct 2026).
- Placed in VLM benchmarks and visual reasoning, one of 16 directions in Multimodal LLMs / vision-language models.
- Builds on 16 of its 93 references that are in the map.
- First posted on arXiv 29 May 2025.

## Placed in

- Multimodal LLMs / vision-language models / [VLM benchmarks and visual reasoning](https://mlsanity.com/mllm/vlm-benchmarks-and-visual-reasoning/)

## Builds on (most cited first, 16 of 16)

- [Qwen2.5-VL Technical Report](https://mlsanity.com/paper/2502.13923/) (2025)
- [Qwen-VL: A Versatile Vision-Language Model for Understanding, Localization, Text Reading, and Beyond](https://mlsanity.com/paper/2308.12966/) (2023)
- [Qwen2 Technical Report](https://mlsanity.com/paper/2407.10671/) (2024)
- [Chameleon: Mixed-Modal Early-Fusion Foundation Models](https://mlsanity.com/paper/2405.09818/) (2024)
- [Pixtral 12B](https://mlsanity.com/paper/2410.07073/) (2024)
- [MLLMs Know Where to Look: Training-free Perception of Small Visual Details with Multimodal LLMs](https://mlsanity.com/paper/2502.17422/) (2025)
- [Forcing Diffuse Distributions out of Language Models](https://mlsanity.com/paper/2404.10859/) (2024)
- [PhD: A ChatGPT-Prompted Visual hallucination Evaluation Dataset](https://mlsanity.com/paper/2403.11116/) (2024)
- [Probing Visual Language Priors in VLMs](https://mlsanity.com/paper/2501.00569/) (2024)
- [IllusionVQA: A Challenging Optical Illusion Dataset for Vision Language Models](https://mlsanity.com/paper/2403.15952/) (2024)
- [GenderBias-VL: Benchmarking Gender Bias in Vision Language Models via Counterfactual Probing](https://mlsanity.com/paper/2407.00600/) (2024)
- [Survey of Social Bias in Vision-Language Models](https://mlsanity.com/paper/2309.14381/) (2023)
- [Heads, Tails, and AI Fails: LLMs, Randomness, and Human Judgments](https://mlsanity.com/paper/2406.00092/) (2024)
- [Investigating Implicit Bias in Large Language Models: A Large-Scale Study of Over 50 LLMs](https://mlsanity.com/paper/2410.12864/) (2024)
- [GPTBIAS: A Comprehensive Framework for Evaluating Bias in Large Language Models](https://mlsanity.com/paper/2312.06315/) (2023)
- [Uncovering Cultural Representation Disparities in Vision-Language Models](https://mlsanity.com/paper/2505.14729/) (2025)

## Built on by (most cited first, 30 of 66)

- [MIRAGE: The Illusion of Visual Understanding](https://mlsanity.com/paper/2603.21687/) (2026)
- [Seed1.8 Model Card: Towards Generalized Real-World Agency](https://mlsanity.com/paper/2603.20633/) (2026)
- [Understanding Language Prior of LVLMs by Contrasting Chain-of-Embedding](https://mlsanity.com/paper/2509.23050/) (2025)
- [Your other Left! Vision-Language Models Fail to Identify Relative Positions in Medical Images](https://mlsanity.com/paper/2508.00549/) (2025)
- [VLMs Need Words: Vision Language Models Ignore Visual Detail In Favor of Semantic Anchors](https://mlsanity.com/paper/2604.02486/) (2026)
- [EdiVal-Agent: An Object-Centric Framework for Automated, Fine-Grained Evaluation of Multi-Turn Editing](https://mlsanity.com/paper/2509.13399/) (2025)
- [Do 3D Large Language Models Really Understand 3D Spatial Relationships?](https://mlsanity.com/paper/2603.23523/) (2026)
- [Video Models Reason Early: Exploiting Plan Commitment for Maze Solving](https://mlsanity.com/paper/2603.30043/) (2026)
- [Do VLMs Perceive or Recall? Probing Visual Perception vs. Memory with Classic Visual Illusions](https://mlsanity.com/paper/2601.22150/) (2026)
- [SPARC: Separating Perception And Reasoning Circuits for Test-time Scaling of VLMs](https://mlsanity.com/paper/2602.06566/) (2026)
- [Same or Not? Enhancing Visual Perception in Vision-Language Models](https://mlsanity.com/paper/2512.23592/) (2025)
- [Mechanisms of Prompt-Induced Hallucination in Vision-Language Models](https://mlsanity.com/paper/2601.05201/) (2026)
- [Watch Before You Answer: Learning from Visually Grounded Post-Training](https://mlsanity.com/paper/2604.05117/) (2026)
- [VisualOverload: Probing Visual Understanding of VLMs in Really Dense Scenes](https://mlsanity.com/paper/2509.25339/) (2025)
- [VISTAQA: Benchmarking Joint Visual Question Answering and Pixel-Level Evidence](https://mlsanity.com/paper/2605.20676/) (2026)
- [PerceptionBench: Evaluating Atomic Visual Perception in Multimodal Large Language Models](https://mlsanity.com/paper/2607.24957/) (2026)
- [Vision Language Models are Confused Tourists](https://mlsanity.com/paper/2511.17004/) (2025)
- [6 Fingers, 1 Kidney: Natural Adversarial Medical Images Reveal Critical Weaknesses of Vision-Language Models](https://mlsanity.com/paper/2512.04238/) (2025)
- [VisRes Bench: On Evaluating the Visual Reasoning Capabilities of VLMs](https://mlsanity.com/paper/2512.21194/) (2025)
- [Mitigating Perceptual Judgment Bias in Multimodal LLM-as-a-Judge via Perceptual Perturbation and Reward Modeling](https://mlsanity.com/paper/2606.02578/) (2026)
- [VLM-SlideEval: Evaluating VLMs on Structured Comprehension and Perturbation Sensitivity in PPT](https://mlsanity.com/paper/2510.22045/) (2025)
- [Do VLMs Read or Rewrite? On Transcription Faithfulness in Vision-Language Models](https://mlsanity.com/paper/2607.21617/) (2026)
- [CAD2DMD-SET: Synthetic Generation Tool of Digital Measurement Device CAD Model Datasets for fine-tuning Large Vision-Language Models](https://mlsanity.com/paper/2508.21732/) (2025)
- [Surgeons Are Indian Males and Speech Therapists Are White Females: Auditing Biases in Vision-Language Models for Healthcare Professionals](https://mlsanity.com/paper/2510.06280/) (2025)
- [MMRPT: MultiModal Reinforcement Pre-Training via Masked Vision-Dependent Reasoning](https://mlsanity.com/paper/2512.07203/) (2025)
- [ArtifactLens: Hundreds of Labels Are Enough for Artifact Detection with VLMs](https://mlsanity.com/paper/2602.09475/) (2026)
- [Stress Tests REVEAL Fragile Temporal and Visual Grounding in Video-Language Models](https://mlsanity.com/paper/2602.11244/) (2026)
- [A Closed-Form Solution for Debiasing Vision-Language Models with Utility Guarantees Across Modalities and Tasks](https://mlsanity.com/paper/2603.12998/) (2026)
- [Beyond Perception Errors: Semantic Fixation in Large Vision-Language Models](https://mlsanity.com/paper/2604.12119/) (2026)
- [Do Physics Foundation Models Learn Generalizable Physics? A Bias-Aware Benchmark Across Physical Regimes and Distribution Shifts](https://mlsanity.com/paper/2605.29283/) (2026)

## Abstract (arXiv)

Large language models (LLMs) memorize a vast amount of prior knowledge from the Internet that helps them on downstream tasks but also may notoriously sway their outputs towards wrong or biased answers. In this work, we test how the knowledge about popular subjects hurt the accuracy of vision language models (VLMs) on standard, objective visual tasks of counting and identification. We find that state-of-the-art VLMs are strongly biased (e.g., unable to recognize the 4th stripe has been added to a 3-stripe Adidas logo) scoring an average of 17.05% accuracy in counting (e.g., counting stripes in an Adidas-like logo) across 7 diverse domains from animals, logos, chess, board games, optical illusions, to patterned grids. Removing image backgrounds nearly doubles accuracy (21.09 percentage points), revealing that contextual visual cues trigger these biased responses. Further analysis of VLMs' reasoning patterns shows that counting accuracy initially rises with thinking tokens, reaching ~40%, before declining with excessive reasoning. Our work presents an interesting failure mode in VLMs and a human-supervised automated framework for testing VLM biases. Code and data are available at: vlmsarebiased.github.io.