Andrew Ng on DeepSeek | Hacker News Reader

Andrew Ng on DeepSeek

deeplearning.ai

55 pointsamrrs8 comments

A decade ago, before the 2017-2022 transformer/LLM revolution, China was ahead in some AI subfields like computer vision and facial recognition.

Top Western AI researchers like Dario Amodei, founder and CEO of Anthropic, used to work for Chinese companies like Baidu a decade ago.

Interestingly some of these people who used to work in the Chinese AI sector a decade ago, like Amodei, are now calling for GPU export restrictions against China. Either they know their moat against China is very limited or they witnessed some things in China that makes them uncomfortable with China getting ahead again in AI.

This just seems like Dario asking for regulatory capture and protection of his company. I don't think you need a deeper explanation than that.

> If the U.S. continues to stymie open source, China will come to dominate this part of the supply chain and many businesses will end up using models that reflect China’s values much more than America’s.

I'm not sure the U.S. as a whole is stymying open source -- or at least there don't seem to be governmental pressures to keep things closed source. Take the llama models, for instance.

Yeah, I am going to blame American/Western companies more than the American/Western governments.

For marketing reasons alone Meta (US) and Mistral (French) should have launched and promoted standalone apps just like DeepSeek did for R1/V3. Many people think DeepSeek's models are the first that can be self-hosted because they simply haven't heard about LLaMA's or Mistral's models.

>I'm not sure the U.S. as a whole is stymying open source

Can these events be considered US is stymying open source [1], [2]?

[1] Navigating Global Regulations and Open Source: US OFAC Sanctions:

https://www.linuxfoundation.org/blog/navigating-global-regul...

[2] Linux Foundation Makes Statement About Banning Russian Programmers (3 Mths Late):

https://news.ycombinator.com/item?id=42899843

Thanks for bringing this to my attention. It’s a bit hard for me to understand the legal intricacies but the point of the article [1] looks like this sentence.

> For developers, this means you need to be cautious about who you interact with and where your contributions come from.

I guess this may prevent USA affiliated developers from collaborating with certain designated foreign developers.

It’s possible that the DeepSeek team or other top foreign developers may become sanctioned by OFAC. Then forking their models may be considered transacting. This would certainly be disastrous for US researchers, if the pace of progress out of foreign teams continues.

But then again I’m not sure why the powers that be would effect such an own-goal.

Wasn't Llama leaked before Meta conceded and made it open source?

Meta was sending LLaMA 1 to anyone with an .edu email. It was always meant to be publicly released.

Reply on news.ycombinator.com