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

AI Is Too Big to Fail

sibylline.dev

Post reply on HN
Print view View original

131–140 of 140 posts

Re: AI Is Too Big to Fail

#131

Post by myko »

jqpabc123 wrote: post #92
miningape wrote: post #38

Earlier quoted context omitted.

Wouldn't the US be cut from the supply if there was a war, regardless of the current tariffs?

The idea from the Chinese side to prevent a war by limiting the ability of the US to produce weapons. The USA and it's allies won WWII mainly due to superior production capacity. This advantage now works in China's favor. And the USA no longer has allies. Bottom line: If China decides to take Taiwan, the USA will have to choose between either doing nothing or mutual annihilation.

Not supporting Taiwan in this conflict would just mean a slow death for the US. Though the US is likely already in that spiral given the trump situation.

Re: AI Is Too Big to Fail

#132

Post by p4ul »

jhallenworld wrote: post #71
Yoric wrote: post #45

Earlier quoted context omitted.

> For Nvidia: all it will take is for one smart grad student to find a better training algorithm to destroy 75% of their value. I don't think so. We've already seen several generations of better training algorithms, but they all rely on CUDA, hence on NVidia.

For inference they already face competition: that deal OpenAI announced with AMD .. at a certain point it will be worth broadening away from CUDA.

Yep, and don't forget about Mojo, which has the potential to make such a transition less painful.

Re: AI Is Too Big to Fail

#133

Post by jqpabc123 »

myko wrote: post #131
jqpabc123 wrote: post #92

Earlier quoted context omitted.

The idea from the Chinese side to prevent a war by limiting the ability of the US to produce weapons. The USA and it's allies won WWII mainly due to superior production capacity. This advantage now works in China's favor. And the USA no longer has allies. Bottom line: If China decides to take Taiwan, the USA will have to choose between either doing nothing or mutual annihilation.

Not supporting Taiwan in this conflict would just mean a slow death for the US. Though the US is likely already in that spiral given the trump situation.

Yes.

It appears that China intends to make this abundantly clear to US "leadership" *before* they launch an invasion --- in hopes of avoiding mutual annihilation which neither side wants.

Re: AI Is Too Big to Fail

#134

Post by laterium »

OptionOfT wrote: post #83

What I really find really weird here is that when I buy stock in Apple, I do so because I think Apple will create new value. But with AI the hope is that the value created will be proportional to wages being removed. I.e. if AI creates more value, it means that the large majority of people will actually have less money, bar a few lucky ones. You have to remember that AI is being touted as the final solution to basica…

Why is the hope that value is proportional to wage being removed and not increased productivity? The problem with this thinking is it already assumes an AGI/ASI replacing all work, rather than a force multiplier like with industrialization, electricity, and computing. AI being touted as the "final solution" doesn't mean anything if AGI is not right around the corner, and it isn't.

If AI is so successful that it can replace even 10% of current work that will be reflected in higher GDP and worker producitivity in the economy. Everyone has this cynical assumption that all benefits will go to the rich, but it hasn't been that way with the previous revolutions either, has it? Did cars only benefit the rich? Trains, radio, television, computers? Initially, that always seems like the case. Who could buy computers in the 80s? But the technology diffuses into the whole economy over time.

Re: AI Is Too Big to Fail

#135

Post by laterium »

robertheadley wrote: post #13

AI isn't going anywhere. It won't magically disappear, but that businesses are trying to use AI in situations that are unsustainable and unneeded.

How do you learn how and when to use AI if you never use it? Just wait another decade to see what others do? Like Kodak and digital photography?

Re: AI Is Too Big to Fail

#136

Post by 7jjjjjjj »

subjectivationx wrote: post #126
tim333 wrote: post #99

Earlier quoted context omitted.

The "~38T national debt bomb" is not caused by AI and not that large compared to debts other countries have dealt with. It's about 100% of gdp and Japan is about 250%. The UK was about 270% after WW2 and that got paid down to 50% before running up a bit. Not ideal but doesn't have to make anything implode.

The problem is Debt to GDP is largely meaningless. It is the total debt that matters. I am sick of reading this bullshit. You don't know what you are talking about and it is just always repeated as a reason to spend more.

If that's true, why do banks ask for my income when I apply for a loan?

Re: AI Is Too Big to Fail

#137

Post by hakfoo »

myko wrote: post #131
jqpabc123 wrote: post #92

Earlier quoted context omitted.

The idea from the Chinese side to prevent a war by limiting the ability of the US to produce weapons. The USA and it's allies won WWII mainly due to superior production capacity. This advantage now works in China's favor. And the USA no longer has allies. Bottom line: If China decides to take Taiwan, the USA will have to choose between either doing nothing or mutual annihilation.

Not supporting Taiwan in this conflict would just mean a slow death for the US. Though the US is likely already in that spiral given the trump situation.

If they can hold on for another 20 years, the PRC could probably have Taiwan without firing a shot.

The US isn't going to get back manufacturing capacity on China's scale, and without real long-term economic plans beyond "bet it all on AGI", what they do have is going to be further strained. They can pump pork into the defense industry specifically, but probably not forever. You can't make a functional economy out of twelve fighter-jet plants and 500,000,000 hectares of unsellable soy.

So over time, you've got a defense partner that's finding it harder and harder to play World Police on a grand scale, and a trading partner that's less and less wealthy and able to dictate terms. Eventually, Taipei will pick up a BRICS application form out of self-interest.

Re: AI Is Too Big to Fail

#138

Post by sshine »

OptionOfT wrote: post #108
sshine wrote: post #107

Earlier quoted context omitted.

> with AI the hope is that the value created will be proportional to wages being removed […] if AI creates more value, it means that the large majority of people will actually have less money, bar a few lucky ones That’s one (not unreasonable) prediction of AI automation. Another is that it will empower the same people whose tasks are made redundant to create value somewhere else. In a way they have to, or they will…

Weirdly enough I have the feeling here that companies don't want to do make more money with the same (retrained) people. They want to make more money with less people: https://www.theregister.com/2025/10/09/mckinsey_ai_monetizat... > All of these copilots are supposed to make work more efficient with fewer people, but my business leaders are also saying they can't reduce head count yet.

> companies don't want to do make more money with the same (retrained) people

Right, a capitalist dream where the factory is fully automated and you can cut all labor cost.

But we're not competing to become laborers at a fixed set of coal mines here.

The service economy is significantly larger than industry; in the US, 77% of GDP. Worldwide, 60%.

There's a limit to how many coffee shops you can have in any one area, but it's limited by purchasing power and not by natural resources that capitalists compete for monopoly over extraction.

There is always something to do.

(Also, I'm grossly overestimating people's willingness and ability to retrain.)

Re: AI Is Too Big to Fail

#139

Post by xg15 »

I know he wanted to end the article with some calls to action that don't sound completely grim, but aren't the last paragraphs a bit contradictory?

1) Build more AI products and do everything imaginable to get AI to produce value - because otherwise, we're doomed.

2) Buy stocks of the tech companies that will most likely be nationalized eventually.

3) Boycott unethical tech companies.

But the companies in (3) are exactly the ones that you'll buy stocks of in (2) and whose services and/or infrastructure you'll build your product on in (1). So how are you supposed to boycott them?

Re: AI Is Too Big to Fail

#140

Post by JohnKemeny »

tonmoy wrote: post #44

Internet was too big to fail in 2000, and in the long run was a net positive for the economy. Yet, a lot of internet companies still have not recovered the bubble bursting

The internet caused social media, which caused the attention economy and trumpism, and will be the bane of the next generations.

It's too soon to tell regarding "net positive for the economy".

Post reply on HN