Comparing Moral Values in Western English-speaking societies and LLMs with Word Associations

TL;DR

This paper investigates the moral values reflected by large language models by analyzing word associations, comparing them with human associations from Western English-speaking communities, revealing systematic differences in moral reasoning.

Contribution

The study introduces a novel method using word associations and seed words from Moral Foundation Theory to compare LLMs' moral values with those of Western English speakers.

Findings

LLMs' moral associations differ systematically from human associations.

A new method propagates moral values through association graphs.

Large datasets of LLM-generated word associations were created.

Abstract

As the impact of large language models increases, understanding the moral values they reflect becomes ever more important. Assessing the nature of moral values as understood by these models via direct prompting is challenging due to potential leakage of human norms into model training data, and their sensitivity to prompt formulation. Instead, we propose to use word associations, which have been shown to reflect moral reasoning in humans, as low-level underlying representations to obtain a more robust picture of LLMs' moral reasoning. We study moral differences in associations from western English-speaking communities and LLMs trained predominantly on English data. First, we create a large dataset of LLM-generated word associations, resembling an existing data set of human word associations. Next, we propose a novel method to propagate moral values based on seed words derived from Moral Foundation Theory through the human and LLM-generated association graphs. Finally, we compare the resulting moral conceptualizations, highlighting detailed but systematic differences between moral values emerging from English speakers and LLM associations.