How Chinese AI Chatbots Censor Themselves

Wait 5 sec.

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.