The Goofus&Gallant Story Corpus for Practical Value Alignment

2025-01-16

AI Summary

This study addresses the AI value alignment challenge by proposing a multimodal social norm modeling framework inspired by developmental moral education paradigms. We introduce the first systematically constructed Goofus & Gallant image–text dataset, explicitly annotated for normative versus non-normative behaviors and aligned with a formal social norm ontology. Data quality is ensured through expert curation, cross-modal consistency verification, and validation against established educational consensus—grounding the dataset in developmental psychology principles. This benchmark fills a critical gap in multimodal, education-informed corpora for value alignment. Empirical evaluation demonstrates substantial improvements in large language models’ performance on social norm judgment tasks: +12.3% accuracy gain and significantly enhanced decision interpretability. The work establishes a novel interdisciplinary paradigm bridging developmental pedagogy and AI alignment research.

Technology Category

• Humans and AI: Learning Human Values and Preferences
• Philosophy and Ethics of AI: Morality & Value-based AI
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems

Abstract

Values or principles are key elements of human society that influence people to behave and function according to an accepted standard set of social rules to maintain social order. As AI systems are becoming ubiquitous in human society, it is a major concern that they could violate these norms or values and potentially cause harm. Thus, to prevent intentional or unintentional harm, AI systems are expected to take actions that align with these principles. Training systems to exhibit this type of behavior is difficult and often requires a specialized dataset. This work presents a multi-modal dataset illustrating normative and non-normative behavior in real-life situations described through natural language and artistic images. This training set contains curated sets of images that are designed to teach young children about social principles. We argue that this is an ideal dataset to use for training socially normative agents given this fact.

Problem

• AI Ethics
• Value Alignment
• Rule Compliance

Innovation

• Goofus and Gallant dataset
• AI ethics training
• Social norm compliance