Cultural Value Alignment in Large Language Models: A Prompt-based Analysis of Schwartz Values in Gemini, ChatGPT, and DeepSeek

This study examines cultural value alignment in large language models (LLMs) byanalyzing how Gemini, ChatGPT, and DeepSeek prioritize values from Schwartz’s valueframework. Using the 40 items Portrait Values Questionnaire, we assessed whetherDeepSeek, trained on Chinese-language data, exhibits distinct value preferencescompared to Western models. Results of a Bayesian ordinal regression model show thatself-transcendence values (e.g., benevolence, universalism) were highly prioritized acrossall models, reflecting a general LLM tendency to emphasize prosocial values. However,DeepSeek uniquely downplayed self-enhancement values (e.g., power, achievement)compared to ChatGPT and Gemini, aligning with collectivist cultural tendencies. Thesefindings suggest that LLMs reflect culturally situated biases rather than a universalethical framework. To address value asymmetries in LLM, we propose multi-perspectivereasoning, self-reflective feedback, and dynamic contextualization. This study contributesto discussions on AI fairness, cultural neutrality, and the need for pluralistic AI alignmentframeworks that integrate diverse moral perspectives.