Perceived (Mis)alignment in generative Artificial Intelligence Varies Across Cultures

The rise of generative artificial intelligence (genAI) has fueled a heated debate about its alignment with human values. Despite widespread use by over 75% of knowledge workers globally, this discourse has been dominated by Western perspectives, often neglecting the values of a diverse user base. In a large pre-registered study with representative samples from six countries (India, Indonesia, Brazil, Germany, U.S., UK; N = 6,362), we examined country-level differences in perceived genAI alignment. People feel genAI falls short of their desired outcomes (d = 1.061). However, perceived misalignment varies across countries (partial η² = 0.132): Western countries are more skeptical and misaligned, whereas Eastern countries are more aligned, optimistic and trusting. These differences cannot be attributed to familiarity with genAI. Country accounts for over 64% of variance, pointing to cultural factors driving these differences. To solve the alignment problem, we must account for the diverse values of global users.