Is It Bad to Work All the Time? Cross-Cultural Evaluation of Social Norm Biases in GPT-4

TL;DR

This paper investigates how GPT-4's understanding of social norms varies across cultures by analyzing its responses to culturally diverse narratives, revealing biases and stereotypes that are often hidden rather than eliminated.

Contribution

It introduces a bottom-up method to evaluate cultural norm biases in GPT-4, highlighting the model's tendency to produce less culture-specific norms and uncovering hidden stereotypes.

Findings

GPT-4 generates less culture-specific norms in narratives.

Stereotypes are hidden rather than eliminated in the model.

Hidden cultural stereotypes can be easily recovered.

Abstract

LLMs have been demonstrated to align with the values of Western or North American cultures. Prior work predominantly showed this effect through leveraging surveys that directly ask (originally people and now also LLMs) about their values. However, it is hard to believe that LLMs would consistently apply those values in real-world scenarios. To address that, we take a bottom-up approach, asking LLMs to reason about cultural norms in narratives from different cultures. We find that GPT-4 tends to generate norms that, while not necessarily incorrect, are significantly less culture-specific. In addition, while it avoids overtly generating stereotypes, the stereotypical representations of certain cultures are merely hidden rather than suppressed in the model, and such stereotypes can be easily recovered. Addressing these challenges is a crucial step towards developing LLMs that fairly serve their diverse user base.