Researchers from Stanford and Princeton found that Chinese AI models are more likely than their Western counterparts to dodge political questions or deliver inaccurate answers.
Več na: https://www.wired.com/story/made-in-china-how-chinese-ai-chatbots-censor-themselves/