Common to Whom? Regional Cultural Commonsense and LLM Bias in India

TL;DR

This paper introduces Indica, a benchmark to evaluate how well large language models understand regional cultural differences within India, revealing significant biases and gaps in regional knowledge.

Contribution

It presents the first regional cultural commonsense benchmark for India, highlighting the variability within a nation and exposing LLM biases and limitations.

Findings

Only 39.4% of questions had agreement across all regions.

LLMs achieve only 13.4%-20.9% accuracy on region-specific questions.

Models show geographic bias, favoring Central and North India.

Abstract

Existing cultural commonsense benchmarks treat nations as monolithic, assuming uniform practices within national boundaries. But does cultural commonsense hold uniformly within a nation, or does it vary at the sub-national level? We introduce Indica, the first benchmark designed to test LLMs' ability to address this question, focusing on India - a nation of 28 states, 8 union territories, and 22 official languages. We collect human-annotated answers from five Indian regions (North, South, East, West, and Central) across 515 questions spanning 8 domains of everyday life, yielding 1,630 region-specific question-answer pairs. Strikingly, only 39.4% of questions elicit agreement across all five regions, demonstrating that cultural commonsense in India is predominantly regional, not national. We evaluate eight state-of-the-art LLMs and find two critical gaps: models achieve only 13.4%-20.9% accuracy on region-specific questions, and they exhibit geographic bias, over-selecting Central and North India as the "default" (selected 30-40% more often than expected) while under-representing East and West. Beyond India, our methodology provides a generalizable framework for evaluating cultural commonsense in any culturally heterogeneous nation, from question design grounded in anthropological taxonomy, to regional data collection, to bias measurement.