Artificial Intelligence and Cultural Context: An Empirical Investigation of Large Language Models’ Performance on Chinese Social Work Professional Standards

Objective: This study examines how well leading Chinese and Western large language models (LLMs) understand and apply Chinese social work principles, testing whether cultural context influences model reasoning and accuracy in non-Western professional settings. Method: Using the Chinese National Social Work Examination covering jurisprudence and applied knowledge, we administered three testing conditions to eight cloud-based LLMs—four Chinese and four Western. We examined responses following the examination’s official guidelines and evaluated the reasoning quality of explanations. Results: Seven models surpassed the 60-point passing threshold, and one model fell slightly below (59.5) in applied knowledge. Chinese models performed stronger on jurisprudence questions (median = 77.0 vs. 70.3 for Western models) but lower on applied knowledge (median = 65.5 vs. 67.0). Models demonstrated strong command of professional terminology but variable performance in culturally specific intervention techniques. Expert review revealed instances where responses reflected societal patterns contradicting professional frameworks, particularly in gender equality scenarios. Conclusions: Both Chinese and Western models demonstrate foundational knowledge of Chinese social work principles. Chinese models excel in regulatory content, while both struggle with culturally nuanced practice scenarios. These findings suggest that technical language ability does not guarantee effective engagement with culturally specific professional content, informing responsible artificial intelligence integration into cross-cultural social work practice.