M$^3$oralBench: A MultiModal Moral Benchmark for LVLMs

2024-12-30

AI Summary

Large Vision-Language Models (LVLMs) lack systematic moral evaluation frameworks for high-stakes domains such as law, finance, and healthcare. Method: This paper introduces MoralBench—the first multimodal moral evaluation benchmark tailored for LVLMs—grounded in the Moral Foundations Theory’s six dimensions. It integrates extended moral scenario texts with paired images generated by SD3.0, supporting three task types: moral judgment, classification, and response generation. MoralBench pioneers cross-modal (text–image) moral assessment, extends Moral Foundations Variables (MFVs) to multimodal settings, and defines an LVLM-specific evaluation paradigm. Contribution/Results: Evaluated on ten state-of-the-art LVLMs, MoralBench reveals significant reasoning deficiencies across fairness, authority, and purity dimensions. The benchmark’s code and dataset are publicly released to foster reproducible, rigorous moral evaluation of multimodal AI systems.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: Language Grounding & Multi-modal NLP
• Computer Vision: Multi-modal Vision

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

Recently, large foundation models, including large language models (LLMs) and large vision-language models (LVLMs), have become essential tools in critical fields such as law, finance, and healthcare. As these models increasingly integrate into our daily life, it is necessary to conduct moral evaluation to ensure that their outputs align with human values and remain within moral boundaries. Previous works primarily focus on LLMs, proposing moral datasets and benchmarks limited to text modality. However, given the rapid development of LVLMs, there is still a lack of multimodal moral evaluation methods. To bridge this gap, we introduce M$^3$oralBench, the first MultiModal Moral Benchmark for LVLMs. M$^3$oralBench expands the everyday moral scenarios in Moral Foundations Vignettes (MFVs) and employs the text-to-image diffusion model, SD3.0, to create corresponding scenario images. It conducts moral evaluation across six moral foundations of Moral Foundations Theory (MFT) and encompasses tasks in moral judgement, moral classification, and moral response, providing a comprehensive assessment of model performance in multimodal moral understanding and reasoning. Extensive experiments on 10 popular open-source and closed-source LVLMs demonstrate that M$^3$oralBench is a challenging benchmark, exposing notable moral limitations in current models. Our benchmark is publicly available.

Problem

• Ethical Judgement
• Multi-modal Models
• Critical Applications

Innovation

• Moral Judgment
• Multimodal Benchmark
• Ethical Reasoning