2025
Culture is Everywhere: A Call for Intentionally Cultural Evaluation
EMNLP 2025
The prevailing “trivia-centered paradigm” for evaluating the cultural alignment of large language models (LLMs) is increasingly inadequate as these models become more advanced and widely deployed. Existing approaches typically reduce culture to static facts or values, testing models via multiple-cho