Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

China’s open-weights AI strategy is winning

werd.io

Post reply on HN
Print view View original

161–170 of 978 posts

Re: China’s open-weights AI strategy is winning

#161

Post by jvanderbot »

Aurornis wrote: post #127
tyleo wrote: post #6

I’m suspicious of some quotes here, “80% of startups using Chinese models,” doesn’t seem quite right to me. I just interviewed at several startups and they were all using the US models. Maybe they have some minor use of Chinese models but the bread-and-butter of most of these businesses model use is the Claude and Codex subscriptions.

It’s a sneaky statistic. You could say that 100% of the startups I worked at used Windows laptops because at least one person had a Windows computer somewhere. If you saw the engineers you’d see 80% Macs and 20% Linux laptops. The statistic would technically be true. I use Chinese open weight models a lot, but they’re not what I reach for when I’m doing important coding work.

I _evaluate_ open weight models all the time. Does this mean I _use_ them? That's a very suspicious statistic.

Re: China’s open-weights AI strategy is winning

#162

Post by WarmWash »

jezzamon wrote: post #69
WarmWash wrote: post #45

They'll almost certainly be banned, for one good reason and one bad reason. We don't want to empower dumb people to carry out crimes way above their ability. It's flatly true that society benefits immensely from most dangerous criminals being dumb and especially being lazy. We're just one "Kid uses free Chinese model to mastermind first ever chemical attack on school" away from society running to slam the "ban" butto…

Kids murdering people at school doesn't tend to change US policies...

Constitutional amendments are way deeper than policies unfortunately. I'm also not aware of a chemical weapons lobby.

Re: China’s open-weights AI strategy is winning

#163

Post by apparent »

tyleo wrote: post #6

I’m suspicious of some quotes here, “80% of startups using Chinese models,” doesn’t seem quite right to me. I just interviewed at several startups and they were all using the US models. Maybe they have some minor use of Chinese models but the bread-and-butter of most of these businesses model use is the Claude and Codex subscriptions.

The full quote:

> When entrepreneurs walk into the offices of Andreessen Horowitz (a16z), a big American venture-capital firm, the odds these days are that their startups are using AI models made in China. “I’d say 80% chance [they are] using a Chinese open-source model,” says Martin Casado, a partner at a16z.

This is very different from what the author portrays. It may be the case that many pre-funded startups are using Chinese open-source models (somewhere in their workflow). But what percent of startups that survive more than a year (either with funding or revenue) are still doing this?

I imagine their pitch is: "look at how well we're doing using open source Chinese models! We'll do even better once we raise money to be able to afford frontier models!"

The way the author presents this quote makes me think he had a preferred narrative and found quotes to back it up. Or he's just a very uncareful reader.

Re: China’s open-weights AI strategy is winning

#164

Post by kirtivr »

furyofantares wrote: post #124
kirtivr wrote: post #117

Earlier quoted context omitted.

Here is a counterexample to whatever counterexample you have in mind - Most popular models on OpenRouter right now: https://openrouter.ai/models?categories=programming&order=mo... Top 7 are all open models.

There very little reason to use OpenRouter to access closed models, instead of just using the provider directly.

One big reason I can think of is to avoid vendor lock-in. This is why most enterprises use a multi-cloud setup, despite the increased complexity.

Re: China’s open-weights AI strategy is winning

#165

Post by q8zd3 »

The article's premise is that USA based LLM providers are losing the AI (cold war) battle because it will not be as adopted as open-weight models, comparing it to closed vs open sourced software. I do not think this is the case because:

* The comparison is weird because open-weight is not the same as open-source software to begin with;

* People based in the USA are at an advantaged position since they have access to both american and chinese models;

* Isn't Running your own model training infrastructure more expansive?

* One can still leverage both, in different phases or use-cases. I do not see how this is an "one or the other" situation.

Re: China’s open-weights AI strategy is winning

#167

Post by altcognito »

lossolo wrote: post #115
slibhb wrote: post #95

Earlier quoted context omitted.

These models are trained to be truthful. Your disagreement isn't with the model, or the US, or the capitalist world but with economics and social sciences. If you want Claude to list arguments for socialism, be explicit about that ("List the best arguments in favor socialism). It will gladly comply. You didn't do that, you asked it to assume a premise that runs contrary to the current state of expert knowledge.

> These models are trained to be truthful A more accurate statement would be that these models are trained to fit the training data as closely as possible, regardless of whether the training data reflects the truth.

Well, yes, but remember there is the reinforcement learning that is applied after, and the system prompts that will bend the results.

Re: China’s open-weights AI strategy is winning

#168

Post by parineum »

verdverm wrote: post #153
parineum wrote: post #132

Earlier quoted context omitted.

Not op but that makes perfect sense to me. "People with mostly bad ideas/execution use Chinese models." Is the point being made. If you slice it to some measure of success, is the statement "Successful start-ups/companies use Chinese models." still true?

How does this comparison sound if we use cloud provider? The Ai is writing code, not executing a startup. The code was never the hard part of startups

Deciding to use open models over closed models is a business decision.

Re: China’s open-weights AI strategy is winning

#169

Post by gs17 »

dyauspitr wrote: post #52
tyleo wrote: post #6

I’m suspicious of some quotes here, “80% of startups using Chinese models,” doesn’t seem quite right to me. I just interviewed at several startups and they were all using the US models. Maybe they have some minor use of Chinese models but the bread-and-butter of most of these businesses model use is the Claude and Codex subscriptions.

What I’ve seen is coding is usually done with US frontier models and anything that is part of a feature on an app and runs at scale on the API is a Chinese model because they are dirt cheap.

That's what I lean towards with the exception that Gemma is also good on a lot of tasks and cheap, while not being Chinese.

Re: China’s open-weights AI strategy is winning

#170

Post by jvanderbot »

ToucanLoucan wrote: post #73
jvanderbot wrote: post #58

Earlier quoted context omitted.

If you believe in some kind of competition-free objective set of market morals, then yes, this is a strange contradiction. If you believe that humans are locked in a productive struggle against each other at the organizational level, and that the knife-edge balance is a feature, not a bug, then it's not so weird to think about. It is simultaneously true that it is in my best interest for prices to sink (as a consumer…

Oh I don't think it's either of those things, I think it's good old fashioned Racism/Xenophobia. We did the same shit to Japan and Korea when they were coming up out of their respective post-war periods, and we still do, to a degree. With China's ruling party also being "communist" (in massive, massive air-quotes) it also lets political actors dust off the McCarthyism to boot. This is, to be clear, not meant as a rin…

So, there's no functional competition or great power struggle or corporate race here? Just racism against chinese people for being chinese? That is your claim specifically?
Post reply on HN