China Ties Open-Source AI To Moral Imperative For Global South
Reactions from ranked influencers
5 posts
«We can’t judge intentions only action. China has signaled their AI strategy clearly while also having their leader tie it to a moral imperative. The cost of a U-turn from here will be high.» Americans may not understand it, but Xi really isn't like Trump. He doesn't backpedal.
Everyone anticipated China defecting from the open-source strategy as they approach the frontier. But just as Kimi 3 come within spitting distance of top US closed models, we’re seeing the Chinese double down on keeping things open. Xi’s speech explicitly makes open access to top models a question of morality, pointing out it will be a fresh historical injustice if the global south are excluded. Ali Baba’s Qwen previously were not open-weight. But the new Qwen 3.8 will be. « Angelica you really expect us to believe the Chinese are doing this out of the goodness of their hearts? » We can’t judge intentions only action. China has signaled their AI strategy clearly while also having their leader tie it to a moral imperative. The cost of a U-turn from here will be high. Meanwhile, if the US doubles down on an almost comically evil plan to stay hegemon forever by locking in an insurmountable AI advantage and monopolizing access. Who are you going to root for as a non-American?
I understand why someone like Bill Bishop can read it approvingly, but it's sad to see more intellectually developed mutuals
I really don't appreciate this genre of "Singaporean anklebiter". Generally, the Chinese from Anglo dominion turn out petty and insufferable. Bro tries to frame Xi's speech as a plea for American tech, and "security" stuff too about export controls. Really? It's about Fable pull.
@henrysgao @deanwball You are one of the most intellectually uninteresting China watchers.
I've heard a few people say that Beijing needs to be AGI-pilled before they become AI safety-pilled. But that's not necessarily the case. Look at the Trump admin's sudden shift on AI safety, not because they were suddenly AGI-pilled but because of more concrete, near-term risks such as cyber. The same can and may be happening now with Beijing. Xi Jinping spoke at WAIC of "preventing loss of control" 防范失控. China is not blind to things like cyber risk. The right question on China and AI safety is not "Does China care?" but "How seriously do the people at the top view these risks relative to other goals, such as promoting AI diffusion across every sector?" Moreover, if there's a sudden shift in risks or risk perception--such as a step-change in model capabilities or some major AI incident or a dramatic policy shift out of Washington--Beijing could move quite swiftly to impose new restrictions. I think others are right to focus on the last lines of Xi's speech as creating a buffer for flexibility. "明者因时而变，知者随事而制。" “The discerning adapt to the times; the wise adjust their measures to circumstances.”
Valuable clarifications on Xi's speech at WAIC He is very concerned about controllability of AI I don't think Xi is an open source fanatic at heart. It can be argued that the Western theory of AI risks is just miscalibrated due to sci-fi priors. He'll keep thinking about it.
Before the World AI Conference, China tried to do a World Internet Conference in 2014. But that it didn't work out so well. Clearly, China has learned some things this time. I really like this annotated commentary on Xi's speech at WAIC by @mattsheehan88: https://mattsheehan.substack.com/p/xi-jinpings-big-ai-speech-annotated