# Jury Duty: Calibration and Orientation Failures in MLLM-as-a-Judge Under Cultural Ambiguity

arXiv: https://arxiv.org/abs/2606.20676 · HTML: https://mlsanity.com/paper/2606.20676/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 31 of its 57 references that are in the map.
- First posted on arXiv 12 Jun 2026.

## Builds on (most cited first, 30 of 31)

- [Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena](https://mlsanity.com/paper/2306.05685/) (2023)
- [Evaluating Large Language Models Trained on Code](https://mlsanity.com/paper/2107.03374/) (2021)
- [CLIPScore: A Reference-free Evaluation Metric for Image Captioning](https://mlsanity.com/paper/2104.08718/) (2021)
- [G-Eval: NLG Evaluation using GPT-4 with Better Human Alignment](https://mlsanity.com/paper/2303.16634/) (2023)
- [CIDEr: Consensus-based Image Description Evaluation](https://mlsanity.com/paper/1411.5726/) (2014)
- [FActScore: Fine-grained Atomic Evaluation of Factual Precision in Long Form Text Generation](https://mlsanity.com/paper/2305.14251/) (2023)
- [Chatbot Arena: An Open Platform for Evaluating LLMs by Human Preference](https://mlsanity.com/paper/2403.04132/) (2024)
- [Large Language Models are not Fair Evaluators](https://mlsanity.com/paper/2305.17926/) (2023)
- [Whose Opinions Do Language Models Reflect?](https://mlsanity.com/paper/2303.17548/) (2023)
- [AlpacaFarm: A Simulation Framework for Methods that Learn from Human Feedback](https://mlsanity.com/paper/2305.14387/) (2023)
- [LLM Evaluators Recognize and Favor Their Own Generations](https://mlsanity.com/paper/2404.13076/) (2024)
- [RLAIF vs. RLHF: Scaling Reinforcement Learning from Human Feedback with AI Feedback](https://mlsanity.com/paper/2309.00267/) (2023)
- [GLM-5: from Vibe Coding to Agentic Engineering](https://mlsanity.com/paper/2602.15763/) (2026)
- [Prometheus: Inducing Fine-grained Evaluation Capability in Language Models](https://mlsanity.com/paper/2310.08491/) (2023)
- [Towards Measuring the Representation of Subjective Global Opinions in Language Models](https://mlsanity.com/paper/2306.16388/) (2023)
- [MLLM-as-a-Judge: Assessing Multimodal LLM-as-a-Judge with Vision-Language Benchmark](https://mlsanity.com/paper/2402.04788/) (2024)
- [JudgeLM: Fine-tuned Large Language Models are Scalable Judges](https://mlsanity.com/paper/2310.17631/) (2023)
- [Humans or LLMs as the Judge? A Study on Judgement Biases](https://mlsanity.com/paper/2402.10669/) (2024)
- [Easily Accessible Text-to-Image Generation Amplifies Demographic Stereotypes at Large Scale](https://mlsanity.com/paper/2211.03759/) (2022)
- [FLASK: Fine-grained Language Model Evaluation based on Alignment Skill Sets](https://mlsanity.com/paper/2307.10928/) (2023)
- [BLEnD: A Benchmark for LLMs on Everyday Knowledge in Diverse Cultures and Languages](https://mlsanity.com/paper/2406.09948/) (2024)
- [Investigating Cultural Alignment of Large Language Models](https://mlsanity.com/paper/2402.13231/) (2024)
- [Photo Aesthetics Ranking Network with Attributes and Content Adaptation](https://mlsanity.com/paper/1606.01621/) (2016)
- [CVQA: Culturally-diverse Multilingual Visual Question Answering Benchmark](https://mlsanity.com/paper/2406.05967/) (2024)
- [Benchmarking Vision Language Models for Cultural Understanding](https://mlsanity.com/paper/2407.10920/) (2024)
- [VLFeedback: A Large-Scale AI Feedback Dataset for Large Vision-Language Models Alignment](https://mlsanity.com/paper/2410.09421/) (2024)
- [CultureVLM: Characterizing and Improving Cultural Understanding of Vision-Language Models for over 100 Countries](https://mlsanity.com/paper/2501.01282/) (2025)
- [Direct Judgement Preference Optimization](https://mlsanity.com/paper/2409.14664/) (2024)
- [FoodieQA: A Multimodal Dataset for Fine-Grained Understanding of Chinese Food Culture](https://mlsanity.com/paper/2406.11030/) (2024)
- [METAL: Towards Multilingual Meta-Evaluation](https://mlsanity.com/paper/2404.01667/) (2024)

## Built on by (most cited first, 1 of 1)

- [Cultural Moment Benchmark: Evaluating Video Cultural Reasoning and Grounding in Southeast Asia](https://mlsanity.com/paper/2608.23065/) (2026)

## Abstract (arXiv)

MLLM-as-a-Judge is conventionally validated by agreement with human annotations, but this metric is undefined when the human pool is culturally heterogeneous. We introduce VOIR DIRE, a multimodal benchmark of 626 culturally paired image--prompt artifacts spanning U.S. and mainland Chinese contexts across food, fashion, and architecture, with annotator pools that are within-pool reliable (a = 0.86/0.74) but cross-pool divergent on evaluation (Q1 r = -0.12). Across six MLLMs, the bias decomposes into two failures: a positivity-floor calibration failure (compressed scale use) and an orientation failure (default to one cultural norm). On this corpus, where contested items are sampled to split the two pools, the floor mechanically validates the more-permissive Chinese reading; persona prompting partially recovers calibration, but the orientation residual survives, evidence the tilt is not reducible to scale compression. Reference-pool in-context demonstrations deepen the orientation residual and inflate the high end rather than restoring use of the low end. Model origin adds a small additive tilt (~0.10 MAE) that is approximately invariant under demonstration. We recommend reporting alignment against each reference pool separately and treating cross-pool divergence as a judge property.