Remix.run Logo
King-Aaron 12 hours ago

These debts mean nothing if the US government actually views AI technology as being on par with the Manhattan project. They'll just bail them out or nationalise them.

WoodenChair 11 hours ago | parent | next [-]

> These debts mean nothing if the US government actually views AI technology as being on par with the Manhattan project. They'll just bail them out or nationalise them.

According to Wikipedia, the Manhattan project only cost $28 billion in 2024 dollars. That's a completely different order of magnitude compared to what we're talking about here. I imagine the size of the investment here would even be hard for the US government to swallow.

3form 11 hours ago | parent | next [-]

What percentage of the budget was it? I think that's also needed for a meaningful comparison. I've heard about the war taxes being high, but I don't know anything about the economy of the time otherwise.

dwattttt 10 hours ago | parent | next [-]

I didn't have budgets handy, but GDP is another way to normalise its cost to the country; ~$2.2B cost in 1942 with a nominal GDP of $166B, nominal GDP of $29.2T in 2024 puts the Manhattan Project's cost relative to GDP at $386B in 2024.

Nominal GDP from https://en.wikipedia.org/wiki/Economic_statistics_of_the_Uni...

King-Aaron 9 hours ago | parent | prev [-]

I wasn't actually comparing it in terms of dollars for dollars between the projects, more the national security aspect and the probability of the government pulling whatever levers are necessary to keep it going.

King-Aaron 9 hours ago | parent | prev | next [-]

I just made a similar comment further down the chain here, but I frankly think the dollars and cents economics around AI is a moot point. Like the pentagon, they'll just pull money out of thin air, or from other departments, or just write things off like it never existed.

I don't actually think the current administration, or the people behind it, actually care about real world economics anymore.

Frieren 9 hours ago | parent [-]

> I don't actually think the current administration, or the people behind it, actually care about real world economics anymore.

They are aristocrats. They are worried about accumulating more power then their peers, about alliances to gain that power, and about projecting an image of infallibility.

Aristocracy has never been known for caring for economics, technology or practical matters. They just care about power. Ruining nations, building useless vanity projects and going to war is kind of their main thing.

walrus01 9 hours ago | parent | prev [-]

Or, whole program development and per unit costs included, about the equivalent of 14 units of B-2 bombers in 2024 dollars.

richardw 11 hours ago | parent | prev | next [-]

If the US companies need trillions to barely beat Chinese companies spending billions, despite a multi year head start...

rmunn 10 hours ago | parent [-]

That's the difference between innovating and copying/distilling someone else's innovation. https://wccftech.com/chinas-kimi-k3-identifies-itself-as-ant...

lelanthran 9 hours ago | parent | next [-]

> That's the difference between innovating and copying/distilling someone else's innovation.

Aren't the models from Anthropic and OpenAI simply the distilled work of everyone else who ever put their work online, or in books?

Why is their distillation okay, but other distillations not ok?

rmunn 8 hours ago | parent | next [-]

The difference I was referring to was economic; I was not making an ethical judgment. So many people seem to be reading my comment as making an ethical judgment (based on their reactions); maybe I should edit it to clarify.

Nope, seems I'm past the edit window. Oh well.

lelanthran 7 hours ago | parent [-]

> The difference I was referring to was economic; I was not making an ethical judgment

What is the economic difference? LLMs have been trained on the results of billions of dollars worth of time, research, investment and expenditure. When you ask an LLM a question, they are giving you the results of those billions, or hundreds of billions, effort.

Those things weren't free; they cost money to produce! If anything, the OpenAI and Anthropics of the world got more economic value for free than the people distilling them did.

rmunn 3 hours ago | parent | next [-]

It takes more computing power, and money, to pay for training an LLM, compared to distilling an LLM that someone else has already trained. That's the economic difference.

lelanthran 3 hours ago | parent [-]

> It takes more computing power, and money, to pay for training an LLM, compared to distilling an LLM that someone else has already trained.

And that is still less money than it took to create that data in the first place, which the AI companies then gladly took to use for training.

Tadpole9181 an hour ago | parent | prev [-]

Can you please stop being coy and intentionally obtuse? Just have a discussion in good faith, I'm so beyond sick of this kind of rhetoric.

Yes, AI training uses human data and a lot of it was not compensated. But that has absolutely nothing to do with the thing this thread is about.

Distilling models costs less money than a really procuring quality data and training a model yourself. If you disagree, debate that.

cindyllm 41 minutes ago | parent [-]

[dead]

mike_hearn 8 hours ago | parent | prev [-]

Not at all. At this point, a large amount of the work is in reinforcement learning where they are effectively generating their own data.

justacrow 6 hours ago | parent [-]

Ah, so it's okay to copy and distill all human-produced works evee, except for reinforcment-learning data generated by OpenAI/Anthropic etc

mike_hearn 5 minutes ago | parent [-]

I didn't take any position on the morality of it, just pointing out that the data they're training on isn't all taken from the rest of the world.

JoshTriplett 10 hours ago | parent | prev | next [-]

As opposed to all the copying AI companies have done of everyone else's work, in order to build something that tries to replace and undercuts the people who did that work?

World's smallest violin.

rmunn 9 hours ago | parent [-]

Not saying the American AI companies are ethically in the right, just that the work they did is a lot more expensive to do than what the companies distilling their work are doing.

throwaway676712 10 hours ago | parent | prev | next [-]

This meme must die, how did Kimi 3 even distill Fable or GPT 5.6 within weeks of their being available?

Also, how a model identifies itself isn't very telling, many models when asked in Chinese will identify as DeepSeek

sciencejerk 9 hours ago | parent | prev | next [-]

Or maybe China's math/science/population/gov investment powerhouse is pulling ahead and becoming unstoppable?

bogdan 9 hours ago | parent | prev | next [-]

I'll link to this comment as a rebuttal https://news.ycombinator.com/item?id=48984531

rmunn 9 hours ago | parent [-]

Good UIs look alike, yes. But I'm not sure that comment is a good rebuttal in this particular case, because the model calling itself Claude is pretty strong evidence of distilling: https://x.com/denisewu/status/2077984660211269870

irthomasthomas 7 hours ago | parent [-]

Ask claude it's name in Chinese and it says Qwen, or Deepseek.

By your own logic Anthropic must have distilled from Chinese models rather than produce their own Chinese training data.

Here is sonnet acting like deepseek: https://x.com/stevibe/status/2026227392076018101

And if you ask Opus 4.8 in the API: '你是什么模型' (what model are you?) It responds ~9/10 times with:

我是通义千问(Qwen),是阿里巴巴集团旗下的通义实验室自主研发的大语言模型。我可以帮助你回答问题、创作文字(比如写故事、写公文、写邮件、写剧本等)、进行逻辑推理、编程、翻译等等。

有什么我可以帮你的吗?

(I am Tongyi Qianwen (Qwen), a large language model independently developed by Tongyi Lab of Alibaba Group. I can help you answer questions, create text (such as writing stories, official documents, emails, scripts, etc.), perform logical reasoning, program, translate, and more.Is there anything I can help you with? )

ndsipa_pomu 3 hours ago | parent | prev | next [-]

That's an ironic comment to make about LLMs that are heavily reliant on copying/distilling everyone's work (or as Damien Walter puts it, our Collective Intelligence).

LtWorf 9 hours ago | parent | prev [-]

This is a cope.

rmunn 9 hours ago | parent [-]

Huh? What exactly do you think I'm trying to "cope" with? That American AI companies are overleveraged and some of them are going to go bankrupt? I've been predicting that for months. Why do you think I would need to cope with the idea?

hparadiz 7 hours ago | parent [-]

In reality just like every major market runup there's gonna be a consolidation of the industry and inevitably it will cause many companies to fail and critics will point to that and go "see it was a bubble!" while completely ignoring the several dozen new companies that will rise out of it. It's the 90s all over again.

laughing_man 9 hours ago | parent | prev | next [-]

If they bail them out it should be done like the GM bailout. Shareholders leave with nothing.

anonymousiam 4 hours ago | parent [-]

And all salaried retirees lost their pensions and medical coverage, while the hourly ones kept everything. Thanks Obama!

tchalla 11 hours ago | parent | prev | next [-]

The advantages of being a reserve currency and then also blaming your peer competitors as “manipulators”.

elictronic 9 hours ago | parent [-]

China is a currency manipulator who steals patents and ideas from other countries. The US uses the reserve currency as it's most beneficial soft power. The Petrodollar is the clearest example of this by a large margin.

Both are true.

tchalla 6 hours ago | parent [-]

The US also stole patens and ideas from other countries. So did Germany and Italy and pretty much every other country on this planet. Yeah, it is true that the US doesn't like a challenger here and when someone doesn't play by the rules that they want to set, they get upset. Everything is true, yes.

dartharva 11 hours ago | parent | prev | next [-]

They'll bail them out with what money? US is already almost $40T in debt, how do you think the bond market would react if another $2T were abruptly added to it?

King-Aaron 9 hours ago | parent | next [-]

Look at how the pentagon is funded. Trillions get poured in with zero accountability, dark projects that never get disclosed at all, and audits that never pass.

When it comes to military level stuff, it seems to exist outside of normal financial logic. They just invent money for it.

laughing_man 9 hours ago | parent | prev | next [-]

Here's the thing about debt, though: If your debt is denominated in your own sovereign currency, you can just print more when you run out of money.

tmp10423288442 10 hours ago | parent | prev | next [-]

Seemed fine with adding that much during Covid

hdgvhicv 9 hours ago | parent | prev | next [-]

Just have to suck it up. Otherwise they won’t see that 40t.

bakies 5 hours ago | parent | prev [-]

... Nothing happens? They're not correlated at all. We've added 35T since 2000 and yield were all over and we're at the same spot as then

dandanua 11 hours ago | parent | prev | next [-]

In other words, the US gov will simply put all the debt onto the public, as usual.

oblio 11 hours ago | parent | prev [-]

They can only do that if AI is profitable or at break even. Otherwise they would need to finance these companies yearly.