A Survey on Moral Foundation Theory and Pre-Trained Language Models: Current Advances and Challenges

2024-09-20

AI & SOCIETY

AI Summary

This study investigates the deep integration of Moral Foundations Theory (MFT) with pre-trained language models (PLMs) to model, evaluate, and interpret multidimensional moral orientations in text. Methodologically, it introduces the first MFT-PLM structured knowledge graph—unifying moral lexicons (e.g., Moral Foundations Dictionary), zero-/few-shot transfer learning, and attention-based interpretability analysis—to systematically assess implicit moral biases in models such as BERT and RoBERTa. Key contributions include: (1) a taxonomy of 12+ prominent MFT-PLM approaches and identification of five recurring limitations; (2) the first evaluation benchmark explicitly designed for moral alignment, coupled with an interpretable improvement pathway; and (3) a paradigm shift from “black-box detection” to “structured alignment” in AI moral modeling, providing both theoretical foundations and a technical framework for building trustworthy, morally aware AI systems.

Technology Category

• Philosophy and Ethics of AI: Philosophical Foundations of AI
• Natural Language Processing: Ethics — Bias, Fairness, Transparency & Privacy
• Machine Learning: Deep Neural Architectures and Foundation Models

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Methods, algorithms and applications for the development of semantic models, knowledge graphs and other forms of structured data models with machine-interpretable semantics
• User Modeling, Personalization and Recommendation: Accountability, Transparency, and Ethics for personalization

Abstract

Moral values have deep roots in early civilizations, codified within norms and laws that regulated societal order and the common good. They play a crucial role in understanding the psychological basis of human behavior and cultural orientation. The moral foundation theory (MFT) is a well-established framework that identifies the core moral foundations underlying the manner in which different cultures shape individual and social lives. Recent advancements in natural language processing, particularly pre-trained language models (PLMs), have enabled the extraction and analysis of moral dimensions from textual data. This survey presents a comprehensive review of MFT-informed PLMs, providing an analysis of moral tendencies in PLMs and their application in the context of MFT. We also review relevant datasets and lexicons and discuss trends, limitations, and future directions. By providing a structured overview of the intersection between PLMs and MFT, this work bridges moral psychology insights within the realm of PLMs, paving the way for further research and development in creating morally aware AI systems.

Problem

• Analyzing moral dimensions in text using Pre-trained Language Models.
• Exploring Moral Foundation Theory's role in cultural behavior understanding.
• Bridging moral psychology and AI for morally aware systems.

Innovation

• MFT-informed PLMs analyze moral dimensions
• PLMs extract moral tendencies from text
• Bridges moral psychology with language models