The problem with this is that Chinese companies are only a few months behind AFAIK. So unless there is some kind of international treaty, implementing a law like this only shifts the problem from “artificial superintelligence is coming* and the US has no laws on the books against its development” to “artificial superintelligence is coming* from China and the US has no effective laws on the books against its deployment**”.
*assuming the worst case scenario where the hype is real and AI technology continues to grow at current pace and becomes a genuine threat, which seems plausible with regards to cybersecurity and job security at least
**assuming that any foreign adversaries such as China or Russia or Iran would have no qualms on using such software against the US, which seems likely
Isn’t much of China’s AI progress coming from distilling US models and making them cheaper? Would they be able to develop its capabilities beyond what we have?
That would be a dangerous assumption. Ofc you can get away saying they’re distilling US models to ease the panic, but it’s an assumption, and would be catastrophic if turned wrong
And for how much I shit on China, the only one actually providing an alternative that isn’t gifting your everything to a private company is some of the Chinese companies. They absolutely have their own intent on why this sophisticated models are open weight, but at the moment that only gives a positive to general people that have the hardware to run those models. And for something like Qwen 3.8 27B, it’s “not much”.
The problem with this is that Chinese companies are only a few months behind AFAIK. So unless there is some kind of international treaty, implementing a law like this only shifts the problem from “artificial superintelligence is coming* and the US has no laws on the books against its development” to “artificial superintelligence is coming* from China and the US has no effective laws on the books against its deployment**”.
*assuming the worst case scenario where the hype is real and AI technology continues to grow at current pace and becomes a genuine threat, which seems plausible with regards to cybersecurity and job security at least
**assuming that any foreign adversaries such as China or Russia or Iran would have no qualms on using such software against the US, which seems likely
Isn’t much of China’s AI progress coming from distilling US models and making them cheaper? Would they be able to develop its capabilities beyond what we have?
That would be a dangerous assumption. Ofc you can get away saying they’re distilling US models to ease the panic, but it’s an assumption, and would be catastrophic if turned wrong
Deepseek publicly claims to have created their models, by distilling competitors and that’s why they’re able to create them really really cheap.
There could be other motives behind such a claim, such as making competitors fear making their most powerful tools public.
Of course they have frontier labs around as well, at the very least at various universities.
True, and even if it is correct there’s no reason they couldn’t start training them without using that shortcut.
And for how much I shit on China, the only one actually providing an alternative that isn’t gifting your everything to a private company is some of the Chinese companies. They absolutely have their own intent on why this sophisticated models are open weight, but at the moment that only gives a positive to general people that have the hardware to run those models. And for something like Qwen 3.8 27B, it’s “not much”.