Representational Harms in LLM-Generated Narratives Against Global Majority Nationalities

TL;DR

This paper investigates how large language models portray diverse national identities, revealing persistent biases, stereotypes, and underrepresentation of Global Majority groups, especially when US cues are present, highlighting urgent ethical concerns.

Contribution

It provides empirical evidence of harmful stereotypes and biases in LLM-generated narratives about non-dominant nationalities, emphasizing the need for culturally aware evaluation methods.

Findings

Minoritized identities are over fifty times more likely to be portrayed in subordinate roles.

US nationality cues amplify representational harms in LLM outputs.

Biases persist even when US cues are replaced with other national identities.

Abstract

Large language models (LLMs) are increasingly used for text generation tasks from everyday use to high-stakes enterprise and government applications, including simulated interviews with asylum seekers. While many works highlight the new potential applications of LLMs, there are risks of LLMs encoding and perpetuating harmful biases about non-dominant communities across the globe. To better evaluate and mitigate such harms, more research examining how LLMs portray diverse individuals is needed. In this work, we study how national origin identities are portrayed by widely-adopted LLMs in response to open-ended narrative generation prompts. Our findings demonstrate the presence of persistent representational harms by national origin, including harmful stereotypes, erasure, and one-dimensional portrayals of Global Majority identities. Minoritized national identities are simultaneously underrepresented in power-neutral stories and overrepresented in subordinated character portrayals, which are over fifty times more likely to appear than dominant portrayals. The degree of harm is amplified when US nationality cues (e.g., ``American'') are present in input prompts. Notably, we find that the harms we identify cannot be explained away via sycophancy, as US-centric biases persist even when replacing US nationality cues with non-US national identities in the prompts. Based on our findings, we call for further exploration of cultural harms in LLMs through methodologies that center Global Majority perspectives and challenge the uncritical adoption of US-based LLMs for the classification, surveillance, and misrepresentation of the majority of our planet.