Skip to main content


China Is About To Pop The AI Bubble


in reply to tracyspcy

TLDW: American companies spend insane amounts of money and have no path towards profitability, and Chinese companies are spending a tiny fraction of that to deliver a product that's mostly as good. On top of that, countries and companies are uncomfortable relying on closed models. So, Chinese providers offering open models at far lower rates mean they're quickly taking over the market. And it turns out that simply having the best model isn't really the most important thing. If a cheaper model can do the work then there's no point paying order of magnitude for a more capable one.
in reply to tracyspcy

And the video points out that the reason AI has trouble getting profitable is because it doesn't just work like how tech "traditionally" works: software pretty much only has development costs, and after that copying and selling infinite copies of the product is pretty much free, where as AI not only has development costs, but it also costs a ton to provide outputs to the customers.
in reply to tracyspcy

No, now that “cognition” of a sort can be smeared on any task at the cost of electricity the future belongs to those who control the generation of electricity.

Local models are much less efficient in terms of both power consumption and cost per token.

I am currently running a local model. It’s not efficient.

in reply to whatiswrongwithyou

Concept of global knowledge available locally sounds plausible to me. I’m not expert in this topic though.
in reply to ☆ Yσɠƚԋσʂ ☆

"They don't have anything for hypergrowth anymore. They don't have the next iphone, the next google search, they don't have anything, they're completely out of ideas" - Ed Zitron
in reply to ☆ Yσɠƚԋσʂ ☆

TLDW:
- China is good at compressing the existing models by stealing the answers they give you to make models that have equivalent performance that are way cheaper (plagiarism)
- shares a bunch of clips of Ed Zitron repeating a lot of the things he said over the last six months about there not being a path to profitability for American AI companies
- the bond market looks similar to how it did back before 2008
in reply to Sparrow_1029

It's worth noting that the whole China is distilling our glorious models narrative is largely nonsense. First of all, everybody distills, so it's not like it's unique to Chinese companies. Second, most of actual research that's published is now coming out of China.
in reply to ☆ Yσɠƚԋσʂ ☆

Making a new cheaper model involves nothing more than taking output from an openathropic model and feeding it into your own. Don't worry about why only Chinese companies are able to do it.
in reply to loathsome dongeater

Obviously people like Scam Altman are just too principled to stoop to the level of these savage Chinese companies.
in reply to Viceversa

I guess that's what I get for talking with dumb people, hey?

You know this post is on ml, right?

in reply to Nonconfrontational

You know this post is on ml, right?


It's not immediately obvious: my client shows posts from all instances of Lemmy.
So I've read the post, considered its content and left the comment. Disregarding the instance of the post.

in reply to Viceversa

It’s not immediately obvious: my client shows posts from all instances of Lemmy.
So I’ve read the post, considered its content and left the comment. Disregarding the instance of the post.


This all started with you saying:

You’re from .ml, of course.
in reply to Chulk

Yes. I read the post and then it clicked to me that the narrative is of a certain... vector.

Then I looked at the author's instance and my suspicions were confirmed.

in reply to Viceversa

Yknow what else your client shows?

::: spoiler With spoilers, you need to click or tap on this text to see what’s hidden beneath!
the instance each post is on
:::

in reply to Rai

This information is not immediately obvious and is rarely noticed because it is rarely useful.
in reply to Viceversa

Seemed suspiciously useful to you just a few comments up.
in reply to Viceversa

That's an amazing counter point, peak .world intellect on display as always.
in reply to Viceversa

Why do ya'll always ad hominem our instance the moment somebody gives a different perspective? Like, honestly... lmfao
in reply to John

It's easier for their world view if they don't see us as actual people with opinions.
in reply to Nonconfrontational

.ml is the only place I have seen so far on Lemmy that regularly removes comments and bans users. ml crowd can't deal with opinions that differ from theirs so just ban and censor them.
in reply to Lucius_Sweet

When the "opinions" is like that bullshit above why not remove it? What the moron is adding to the discussion?
in reply to ZeroHora

I'm sure the irony of responding to a bullshit comment with a bullshit comment of your own is lost on you. Your comment is a great example of the mentality of those who are pro censorship.

I would suggest learning basic grammar before responding to anyone else.

What the moron is adding to the discussion?


My 8 year old can write at a more intelligible sentence than this. I do not trust the opinions of those with an inferior intellect to him.

in reply to Lucius_Sweet

I don't know if your brain can process this but the world is a little bit bigger than the "western world" and not everyone in this site has English as the main language.

Pretty sure your 8 year old(son/cat/niece/nephew/whatever) can understand what I wrote.

in reply to ZeroHora

I understand perfectly well, I am from Europe, my country has 2 recognised languages.

I showed my 8 year old both of your comments, he had a good laugh when I told him an adult wrote them.

in reply to Lucius_Sweet

Maybe you should try teaching them empathy and understanding instead of your shitty judgemental attitude.
in reply to Nonconfrontational

I'm afraid I am fresh out of empathy and understanding when it comes to the illiterate attempting to insult my intelligence.
in reply to Lucius_Sweet

Then you should feel even stupider for judging an ESL person on their grammar.
in reply to Nonconfrontational

I don't know why you would think that because I definitely do not. They have even less of an excuse now that grammar and spelling checking are freely available. If the comment was intelligible I would have had some sympathy, I have no sympathy for slop.

I hate this modern world where people expect medals for trying.

in reply to Lucius_Sweet

I get banned from .world communities like once a week lol. My comments are sometimes removed for civility, but more often than not the mod just doesn't like to hear America get trash talked.
in reply to Nonconfrontational

I was banned and had comments removed for suggesting that China was far from a socialist country and have just figured out a different approach to exploiting workers surplus labor.

Seems like this is a common theme for power tripping mods as well as any other system of power. Proves the main point I was trying to make about vertical power structures being the problem. Give most people any degree of power and it goes to their heads. Our monkey brains can't handle power.

in reply to Lucius_Sweet

I mean.. China is a provably socialist system, but I agree power generally corrupts.
in reply to Lucius_Sweet

You are not owed a platform for your misinformation.

Your entire comment history is just you being a fed. Of course the anti-fed instance is not gonna allow that lol

in reply to BlameTheAntifa

Are we playing with tags now? Ok
This entry was edited (1 week ago)
in reply to Sparrow_1029

China is good at compressing the existing models by stealing the answers they give you to make models that have equivalent performance that are way cheaper (plagiarism)


Lol plagerism is when they take your shitty product and actually make it better apparently. Begs the question why China manages to succeed at this when Western companies that own the original product can't. You'd think they'd always be one step ahead of those damn plagerising Chinese scum or at least would be able to do the exact same thing China does.

Also if you're calling plagerism on this I'm calling plagerism on all of AI. You can't accept any generative AI while simultaneously pearl clutching when someone else plagerises your country's AI. So pick one. Either we ban all AI for plagerising all of humanity or we don't give a shit when one AI company plagerises another.

This entry was edited (1 week ago)
in reply to HiddenLayer555

I didn't really intend "plagiarism" to convey my opinion (which is that the AI companies can get fucked), but more as a neutral term to mean Chinese companies are copying to make cheaper models, as asserted by the video
This entry was edited (1 week ago)
in reply to Sparrow_1029

Fair enough, I apologise for jumping to conclusions then.
This entry was edited (1 week ago)