Pre-Trained Language Models Represent Some Geographic Populations Better Than Others

TL;DR

This study evaluates how well pre-trained language models from OPT and BLOOM represent diverse global populations, revealing significant geographic biases favoring US and UK populations over South and Southeast Asia.

Contribution

It introduces a spatial probing method to quantify geographic representation biases in large language models, highlighting uneven global coverage.

Findings

Models perform better for US and UK populations.

Models poorly represent South and Southeast Asian populations.

Biases are consistent across different model families.

Abstract

This paper measures the skew in how well two families of LLMs represent diverse geographic populations. A spatial probing task is used with geo-referenced corpora to measure the degree to which pre-trained language models from the OPT and BLOOM series represent diverse populations around the world. Results show that these models perform much better for some populations than others. In particular, populations across the US and the UK are represented quite well while those in South and Southeast Asia are poorly represented. Analysis shows that both families of models largely share the same skew across populations. At the same time, this skew cannot be fully explained by sociolinguistic factors, economic factors, or geographic factors. The basic conclusion from this analysis is that pre-trained models do not equally represent the world's population: there is a strong skew towards specific geographic populations. This finding challenges the idea that a single model can be used for all populations.