Culturally Prompted Large Language Models Remain Anchored to Western Values

One proposed solution for reducing cultural bias in large language models (LLMs) involves “cultural prompting,” in which LLMs are explicitly instructed to respond on behalf of a cultural group. However, past studies have yielded inconsistent claims about the efficacy of cultural prompting. Here we t