A Cross-Cultural Comparison of LLM-based Public Opinion Simulation: Evaluating Chinese and U.S. Models on Diverse Societies

The emergence of powerful large language models (LLMs) from non-Western cultural contexts, such as China's DeepSeek, prompts a critical question: Does an LLM's cultural origin influence its ability to simulate public opinion across different societies? To address this question, this study evaluates DeepSeek's ability to simulate public opinion in comparison to LLMs developed by major U.S. and Chinese tech companies. Utilizing survey data from the American National Election Studies (ANES) and China's Zuobiao dataset, we assess how these models predict public opinions on key social issues in both nations. Our findings indicate that an LLM's cultural origin does not confer a straightforward “home-field advantage.” Instead, all evaluated LLMs, regardless of origin, exhibit significant cultural and demographic biases, struggling to capture nuanced perspectives within specific societal groups. For example, DeepSeek-V3 performs best in simulating U.S. opinions on the abortion issue compared to other topics such as climate change, gun control, immigration, and services for same-sex couples, primarily because it more accurately simulates responses when provided with Democratic or liberal personas. For Chinese samples, DeepSeek-V3 performs best in simulating opinions on foreign aid and individualism but shows limitations in modeling views on capitalism, particularly failing to capture the stances of low-income and non-collegeeducated individuals. It does not exhibit significant differences from other models in simulating opinions on traditionalism and the free market. Ultimately, all LLMs exhibit the tendency to overgeneralize a single perspective within demographic groups, often defaulting to consistent responses. These findings highlight the need to mitigate cultural and demographic biases in LLMdriven public opinion modeling, calling for approaches such as more inclusive training methodologies to ensure LLMs can accurately and fairly represent diverse global viewpoints.