Chinese AI models are politically complex, not just government mouthpieces, study finds
🤖 AI-generated content — The title and summary were produced automatically by artificial intelligence, without human editorial review.
A study by AI benchmark firm Vals AI tested 25 models on political questions regarding China, finding that Chinese models respond less uniformly than expected and some Western models occasionally adopt Beijing's official positions. The research measured alignment scores ranging from −1 to +1 and showed that prompt language heavily influences responses, with Simplified Chinese yielding more pro-government answers than English, Russian, or Spanish. Individual models varied significantly, with GLM-5.3 Flash explicitly confirming China's documented human rights violations.
Continue in the app — vote & join in ➔Source: The Decoder (AI) · via ahirlevel.hu