Richer Output for Richer Countries: Uncovering Geographical Disparities in Generated Stories and Travel Recommendations

2024-11-11

arXiv.org

AI Summary

This study provides the first empirical evidence of systematic geographic bias in large language models (LLMs) for geospatial tasks. To quantify such bias, we introduce a cross-national, multi-task evaluation framework—grounded in 100K travel requests and 200K geographically anchored stories—and integrate geographic entity recognition, sentiment analysis, n-gram diversity metrics, and rigorous statistical significance testing (bootstrap/permutation). Results reveal substantial disparities: travel recommendations for low-income countries exhibit 37% lower uniqueness and 52% fewer location mentions; generated stories contain 2.1× higher frequency of hardship-related negative sentiment terms. Beyond establishing a quantifiable link between geographic knowledge representation and societal impact, our work proposes the first scalable, structural framework for assessing geographic bias in LLMs—thereby exposing an implicit geographic digital divide.

Technology Category

• Natural Language Processing: Ethics — Bias, Fairness, Transparency & Privacy
• Machine Learning: Large Multimodal Models (LMMs)
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Large language models for search

Abstract

While a large body of work inspects language models for biases concerning gender, race, occupation and religion, biases of geographical nature are relatively less explored. Some recent studies benchmark the degree to which large language models encode geospatial knowledge. However, the impact of the encoded geographical knowledge (or lack thereof) on real-world applications has not been documented. In this work, we examine large language models for two common scenarios that require geographical knowledge: (a) travel recommendations and (b) geo-anchored story generation. Specifically, we study five popular language models, and across about $100$K travel requests, and $200$K story generations, we observe that travel recommendations corresponding to poorer countries are less unique with fewer location references, and stories from these regions more often convey emotions of hardship and sadness compared to those from wealthier nations.

Problem

• Geographical biases in language models
• Impact on travel recommendations
• Emotional disparity in story generation

Innovation

• Analyzes geographical biases in language models
• Compares travel recommendations across countries
• Examines emotion in geo-anchored story generation