Arcee AI @stochasticchasm argues Chinese open-source models are not inherently dangerous: "I don't think there's any form of malicious intent encoded in Chinese models." "With closed APIs, you don't know if the model changes day to day. I mean, I'd expect not, but you see people always complain, oh, this feels quantized today. I don't believe that's actually happening. But the fact that people make these comments shows that people value predictability a lot." "That's a good benefit of…