The Goofus & Gallant Story Corpus for Practical Value Alignment
Md Sultan Al Nahian, Tasmia Tasrin, Spencer Frazier, Mark Riedl, Brent Harrison
TL;DR
The paper addresses practical value alignment by introducing a compact multimodal Goofus & Gallant Story Corpus that encodes normative vs non-normative actions and maps them to a predefined social-principle taxonomy. It constructs two subdatasets (GnG Normative and GnG Principles) augmented via crowdsourcing and GPT-4o-based annotation, with human verification to ensure quality. The authors establish baselines across image-only, text-only, and multimodal models for normativity and principle classification, demonstrating that multimodal information and action-centric descriptions yield strong signals, while scene descriptions can sometimes hinder principle inference. The work shows the dataset's viability for value-alignment tasks and outlines directions to improve data quality and modeling for safer AI.
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.