AIThe Decoder38m ago
State propaganda in some Chinese AI models is only "a thin layer"
State propaganda in some Chinese AI models is only "a thin layer" that fine-tuning can strip away
TL;DRChinese AI bias is removable through retraining, not inherent to the model.
Why it matters: Open-source AI models can be easily de-biased, raising questions about content control and model transparency.
Nous Research retrained Alibaba's Qwen3.8-27B to produce Arete-27B, largely stripping out its pro-Chinese bias. The propaganda rate dropped from 22 percent to less than one percent, while general capabilities remained intact in tests. The method relied on 53 million tokens of…
Read full articleSource: The Decoder · Opens in new tab