OpenAI CEO Sam Altman says that export controls alone won’t hold back China's AI ambitions — “My instinct is that doesn’t work”

zsydeepsky said:

having China provide open-source AI models is a good thing for everyone, including Americans.

Most of their models aren't and won't be open source. Deep Seek's open source model was mostly a publicity stunt, and it worked!

If China is so committed to open source models, let's have them prove it by open sourcing the Tik Tok algorithm.

It's okay, China can burn as much energy for nothing as they want.
I actually won't be surprised if "AI" hype ambitions will bring China down eventually.

acadia11 said:

The US needs to realize they can not hold back China's AI ambition, it's trying to hold back water, the wheel, electicity, internet , railroads, ... China unlike say Russia has made being the worlds technological juggernaut their priority. Remember historically China was a cradle of man's technological innovation ... after a few centuries of isolationist ideals put them in the backseat and they've been on this March since opening their borders again the 1800's and really made this a priority in the 1970's or being in the position as they turned away from egalitarian ideals of peasent workers and what not. The truth is China is just re-asserting their place in the world which has historically been one of the world's super powers ... it's always been the most populous area since man walked out of Africa and the fact is advancement in culture has generally been a function of population size. China simply put has too many resources and too many minds and access to information that stopping them from AI advancement is a silly idea. Jensen is correct on one thing and it makes sense he understood Nvidias power was in essentially becoming the framework for AI by supplying the compute and building CUDA to take advantage of the GPU compute.

NOw that's ubiquitous Nvidia is well Nvidia much like x86 (IBM compaitible) won the PC race. Mkaing China reliant on US technology was the only real answer but China's government understnads the importance of self sufficiency controlling exports will just accelerate them building out a complete full stack solution that rivals Nvidia, and with the current geo-political landscape their will be plenty of adopters of Chinese tech. YOu see it especially in Africa. While the US is going into an isolationist mentality ... they are creating a vacuum and China more so then say Russia is filling that void. China is number two in AI only behind the US , and they'll ahve the compute capabilities in next couple of years and full stack framework to go with it with in that timeframe. But I believe the idea is to get the best deal possible now on a clearly soon to be much more formidable frenemy, foe, friend , enemy ... don't know o the world stage.

I don't even know where to begin with this bit of Xinhua propaganda...
First of all, it was the Roman empire --not China, that was the ancient superpower, especially in matters related to tech innovation (I don't care what that weak, poorly-cited article in Wikipedia has to say on the matter). But the real defining historical moment was the Western Renaissance, which brought forth scientific inquiry and innovation like the Cambrian explosion brought forth life: great names like Newton, Copernicus, Goethe, Jenner, Hooke, ... They all march past in an endless procession. In short order, modern chemistry, mathematics, physics, astronomy, biology, they were all borne of Western thought. China was quite aware of our achievements, yet did nothing but invent gunpowder by accident, and "invent" paper of all things. Until the 21st century, this is the essence of China's contribution to the world. The UK during the 19th century of humiliation failed to force China into Renaissancing. Hell, even the United States's mighty, towering scientific achievements post WW2 failed to make China Renaissance. Nixon, and later American industrialists and universities had to go down there and shove a love of science down the throats of the Chinese, and even then they didn't quite grab hold of it until they started making big money from it.

Chairman Xi made China do the 180* turn around with his talk of national rejuvenation. Jinping has been speaking of global domination by 2049 (guess the significance of the year. The year is no accident). This has been resonating with the Chinese people, making them actually believe in all this poppycock about how China somehow has a right to Lord over the rest of the world because they're the middle kingdom or some nonsense. Nevertheless, the Chinese have been lapping it up, working hard to meet Chairman Xi 's 2049 unspoken deadline.

There are currently 23 years to go. Do you guys think China can kick our asses by then? That means, of course, that China's GDP has to blast past ours, and so far, China is doing a poor job of it. We sit at $30T and China at only $19T.

It's flawed thinking to say that people are an asset. They are, up to a point. When you get into populations of the scale of China's, then their numbers become a liability because you just end up with a surplus number of engineers sitting around twiddling their collective thumbs, waiting for an innovation to arrive so that they can engineer products of it. Unlike infrastructure development--which can take place at breakneck speed (and whose speed is population-dependent), science really is zero-sum: s scientific development can only accelerate to a point, and this point has been reached over 30 years ago. You notice how we're still using cellphones? They were introduced in 2007 and have undergone little change since then. They will continue to dominate for the foreseeable future because cellphone science depends on other branches of science for its progress and ultimate retirement. Soon, cellphones will have been around longer than land-line telephony with autorouting, and have already been around longer than telephones with push-button dialing. All of this stagnation despite a near 20-fold increase in the number of engineers since the late 20th century.

China is beating us in patents but the whole world by now knows they are shit patents, and that the United States remains the hub of scientific advancements. You ever read a journal paper by Chinese authors? These biased nationalists will only work with other Chinese, cute other Chinese in their work, and be financed by Chinese institutions. Therefore, Chinese patents or journals cannot be taken seriously, nor are they taken seriously.

China's great numbers are helpful only in areas of manufacturing and infrastructure building. When it comes to the scientific pursuit, only a curious mind and a passion for the work will meet the overbearing and demanding finish line; sending a bunch of kids through engineering college in a regimented, hierarchical and authoritarian society won't cut the mustard.

You guys ever notice that the great Chinese ideas of the 20th and 21st centuries have been accomplished by Chinese scientists who were educated at American universities? There's a reason for this, and that reason is not related to educational rigor!

bit_user said:

Most of their models aren't and won't be open source. Deep Seek's open source model was mostly a publicity stunt, and it worked!

If China is so committed to open source models, let's have them prove it by open sourcing the Tik Tok algorithm.

Good point about tic toc. Also, the primary and only reason why DeepSeek and other Chinese LLMs are open-source is because... Wait for it... It's because the Chinese AI is largely based off of American AI. No big risk to proprietary company Intel if that Intel isn't even original...

bit_user said:

Most of their models aren't and won't be open source. Deep Seek's open source model was mostly a publicity stunt, and it worked!

If China is so committed to open source models, let's have them prove it by open sourcing the Tik Tok algorithm.

I see you haven't been following the AI model updates.

You can check all open-source AI models from Huggingface, the Chinese models occupy more than half of the first page of the trending models. I'll list them for you to check yourself:

Qwen/Qwen-Image-Editdeepseek-ai/DeepSeek-V3.1-Basetencent/Hunyuan-GameCraft-1.0zai-org/GLM-4.5VAIDC-AI/Ovis2.5-9Bopenbmb/MiniCPM-V-4stepfun-ai/NextStep-1-Largerednote-hilab/dots.ocrPhr00t/WAN2.2-14B-Rapid-AllInOne
Then some models derived from them, such as:
janhq/Jan-v1-4Blightx2v/Qwen-Image-Lightning
The thing is, Open-source models are almost dominated by Chinese models now. And they are not just listed models, but actually have been used, not only by Chinese consumers, but also by Western users. You can see it from OpenRouters model usage ranklists. in this list, you can see other Chinese model names (since older models might still be used in some existing systems), such as:

ERNIE 4.5 21B A3BQwQ 32B RpR v1GLM Z1 32BInternVL3 14BMiniMax M1GLM 4 32BQwen-Turbo
On the other hand, American companies have almost fully retreated from open-source models, like Meta discontinued Llama, Google open-sourced only Gemma but not Gemini, Elon Musk promised to open-source Grok2 which hasn't been realized yet, and Claude that never went open source.

The Chinese models are gaining traffic even from American users, literally because they have very little choice. Only China is providing open-source models that match cutting-edge private models with vast variety of choices.

So no, DeepSeek is not a publicity stunt, it's just the first name you saw on mainstream news reports, and it's only a glimpse of China's AI industry.

At best you slow them down a bit. But while you do that, they develop independent means and the means to compete in the global hardware market. Once you start using export controls, you have the beginnings of annoying defeat and lacking the confidence to "win".