Global PIQA: Evaluating Physical Commonsense Reasoning Across 100+ Languages and Cultures

TL;DR

Global PIQA is a comprehensive, culturally-aware benchmark for evaluating physical commonsense reasoning across over 100 languages and cultures, revealing disparities in LLM performance especially in low-resource languages.

Contribution

It introduces a large-scale, culturally-diverse benchmark for multilingual commonsense reasoning, constructed by a global community of researchers, highlighting cultural and resource-based performance gaps in LLMs.

Findings

LLMs perform well overall but struggle with low-resource languages.

Open models generally underperform compared to proprietary models.

Many languages and cultures still lack sufficient everyday knowledge in LLMs.

Abstract

To date, there exist almost no culturally-specific evaluation benchmarks for large language models (LLMs) that cover a large number of languages and cultures. In this paper, we present Global PIQA, a participatory commonsense reasoning benchmark for over 100 languages, constructed by hand by 335 researchers from 65 countries around the world. The 116 language varieties in Global PIQA cover five continents, 14 language families, and 23 writing systems. In the non-parallel split of Global PIQA, over 50% of examples reference local foods, customs, traditions, or other culturally-specific elements. We find that state-of-the-art LLMs perform well on Global PIQA in aggregate, but they exhibit weaker performance in lower-resource languages (up to a 37% accuracy gap, despite random chance at 50%). Open models generally perform worse than proprietary models. Global PIQA highlights that in many languages and cultures, everyday knowledge remains an area for improvement, alongside more widely-discussed capabilities such as complex reasoning and expert knowledge. Beyond its uses for LLM evaluation, we hope that Global PIQA provides a glimpse into the wide diversity of cultures in which human language is embedded.