I don't think most people in the west realize how structurally hostile AI safety is for Chinese researchers. There's so many orgs that have hawkish content that frame China as an existential adversary such that it's a reputational risk for many Chinese people to engage. I have to think twice about posting stuff here.
Some of the current efforts are trying to create a tiny, well-defined "safe space" for Chinese to engage, which is of course helpful, but that is bottlenecked by org capacity. Why can't people separate out the hawkish stuff and do the reverse, such that most places are safe for Chinese to engage?
To be clear, the original content was meant to just share how I feel, not intended as mockery, and I just realised that it might come across that way.
For social, commercial, and political reasons, we can't put everything online, which makes the vast majority of safety work invisible to the west, that doesn't mean no one working on AI safety in China. Charbel-Raphaël also mentioned invisible work in AI governance.
Xi's speech included "ensure AI stays under human control". TC260 AI Safety Governance framework 3.0 lays the groundwork for future compulsory ... (read more)