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

71–80 of 140 posts

Re: AI Is Too Big to Fail

#71

Post by jhallenworld »

Yoric wrote: post #45
jhallenworld wrote: post #43

>~75–80% of S&P 500 gains Dot com bust was 50% of S&P / 80% on Nasdaq... we survived it.. For Nvidia: all it will take is for one smart grad student to find a better training algorithm to destroy 75% of their value. Why I think a better algorithm is out there: Total installed GPU Tflops = 4 billion. Tflops of human brain = 1 million

> 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.

Re: AI Is Too Big to Fail

#72

Post by fisf »

devin wrote: post #25
philipwhiuk wrote: post #16

Earlier quoted context omitted.

Anti-establishment is sort of a requirement for doing a start-up. Aggressiveness is a requirement for doing a start-up in times of constrained capitalism Low quality is driven the by the availability of capital for dumb ideas. So... a while now?

These startups backed by venture funds are anti-establishment? Are you joking?

People like to think that they are anti-establishment and disruptive at a startup. They conveniently ignore who is writing the cheques.

Re: AI Is Too Big to Fail

#73

Post by miningape »

travisgriggs wrote: post #49
philipwhiuk wrote: post #18

Earlier quoted context omitted.

Pensioners and retirees willingly dumping money into index funds without oversight are the dumb money that enables all of this. So yes, they deserve a haircut.

This is pretty myopic, or something. Shows, at least, a real ignorance about the available possibilities (or lack thereof)—at scale—to the common worker for “saving for a later day”. But I’m all ears. Now that you’ve diagnosed how 401K investing fools get what they deserve, care to offer any alternative solutions as to the entire work force should have been saving towards retirement.

(kinda unrelated but) I personally hate these forced savings schemes from the government. At least in my country the rates are low so it almost feels like I'm donating the entire difference in rates between the pension scheme and the S&P500 + taxes straight to the government.

At least give those of us brave (or stupid) enough to do something different with the money the option to.

Re: AI Is Too Big to Fail

#74

Post by LikeBeans »

I'm skeptical too however I think AI can add value by increasing knowledge worker productivity. Maybe save some money by needing less staff for some tasks. Maybe to mine vast amounts of data looking for a pattern. So I think it is useful however my skepticism is about the cost vs reward. Is it really worth the amount of money pumped into it, not sure. Time will tell.

Re: AI Is Too Big to Fail

#75

Post by mcv »

Yoric wrote: post #59
mcv wrote: post #53

I think this is mostly about the US, right? Most countries haven't invested all that much in AI. Even if all this new AI would completely vanish overnight, I'm pretty sure most of the world wouldn't care one bit. Of course many AI companies and investors would be bankrupt; they'd be the real losers here. Many other companies and organisations would have wasted a lot of money on nothing, but they'll write it off and m…

Well, last few times the US had a big recession, they took down the whole world with them. The biggest recession led almost directly to WWII, and that was before the US centralized that much power or money. I don't live in the US, but I'm rather worried.

That's true. But thanks to Trump's tariffs, much of the world is working to make itself less dependent on the US. There's talk that the Euro could become the new reserve currency. Maybe the Renmibi.

It's quite possible that the American Century will end with a massive crash, and outside of the US, Europe will probably suffer quite a bit of damage too if we don't detach ourselves more from the US, but I still think most of the world will be fine. Though I will admit that could be partially wishful thinking.

Re: AI Is Too Big to Fail

#76

Post by marcosdumay »

jcranmer wrote: post #41

The idea that the Republicans are going to bail out AI and pay for it with massive tax raises on the plebs is just utter bonkers, if for no other reason than they're currently utterly incapable of getting any legislation passed, let alone one so deeply unpopular as that.

Their government is currently closed, but that BBB thing passed... So, I have no idea.

Re: AI Is Too Big to Fail

#77

Post by anon84873628 »

catigula wrote: post #28

This talk of arms race with China is very, very dangerous. We need to start having a more important conversation if people really believe this because, otherwise, why shouldn't China just kinetically strike US infrastructure and vice versa? If AI is an executioner's blow to your opponent by enabling you to dominate them in terms of hacking, economy, government, politics, etc. kinetic strikes are more than justified.…

I don't even understand what people think we're racing towards.

What the global masses want are good food, nice houses, fast transportation, TVs and cell phones... Physical comforts.

It doesn't matter what a superintelligence dreams up if you can't physically produce any of it. As the article says, China is the only place that actually has the manufacturing and robotics capability.

In the US we maybe have a hope of seeing more miracle healthcare, marginally more optimized industrial agriculture, and I guess even more lethal military technology. At best it can design all the factories and energy plants we are missing.

And putting all the white collar knowledge workers out of work- what would that accomplish? People aren't ready for the massive social reorganization that this tech would imply. Would it even survive the revolution it causes??

Re: AI Is Too Big to Fail

#78

Post by marcosdumay »

Fernicia wrote: post #14
terandle wrote: post #8

> Think of it like WWII, only only instead of planting victory gardens to beat the Nazis, we're building AI apps and finding ways to create the economic value needed to cover the reckless bets being made by the elites. LOL fuck this, the stock market deserves to burn to the ground.

When did the HackerNews comment section turn into this? Low quality, aggressive, fervently anti-establishment.

Have you looked at the establishment recently?

Re: AI Is Too Big to Fail

#79

Post by b-karl »

So maybe we are talking too much about AI as a tech bubble (like the dotcom bubble) and it is actually closer to an arms race like the nuclear programs or space race? For better and for worse there are a lot of secondary value and ventures created from those and one could see AI ventures in a similar light.

A key difference is of course, at least in the US, that the research is led by corporations as opposed to research institutes and government.

Re: AI Is Too Big to Fail

#80

Post by jcranmer »

wrs wrote: post #67
jcranmer wrote: post #41

The idea that the Republicans are going to bail out AI and pay for it with massive tax raises on the plebs is just utter bonkers, if for no other reason than they're currently utterly incapable of getting any legislation passed, let alone one so deeply unpopular as that.

What? Did you miss the BBB? They’ll pass any legislation they’re told to.

The BBB, i.e., the bill passed via the special mechanism that only works once a year and has stringent requirements on what can be in the bill for the mechanism to work.
Post reply on HN