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

After the AI boom: what might we be left with?

blog.robbowley.net

Post reply on HN
Print view View original

121–130 of 489 posts

Re: After the AI boom: what might we be left with?

#121

Post by mikert89 »

cactusplant7374 wrote: post #53
mikert89 wrote: post #8

Earlier quoted context omitted.

dude again, we have computers that can talk and make decisions. We have birthed something here. We have something, this is big.

You should post something more substantial than this.

its a problem of imagination, "situational awareness". People are simply not aware of what we have discovered, their minds cannot see beyond a chatbox. thats not even to mention the smoothness of the loss functions the big ai labs are seeing, the smooth progression of the scaling laws. its all there, it progresses daily

Re: After the AI boom: what might we be left with?

#122

Post by pclmulqdq »

trenchpilgrim wrote: post #88
nemomarx wrote: post #86

Earlier quoted context omitted.

After heavy use, though? I don't think they mean aging out of being cutting edge but actually starting to fail sooner after being used in DCs.

They're mostly solid state parts. The parts that do wear out like fans are easily replaced by hobbyists.

This is not correct. Solid-state parts wear out like mechanical parts, especially when you run them hot. The mechanism of this wear-out comes from things like electromigration of materials and high-energy electrons literally knocking atoms out of place.

Re: After the AI boom: what might we be left with?

#123

Post by cyberax »

flyinglizard wrote: post #10

I admit to only being in this industry for three decades now, and only designing and implementing the thermal/power control algo of an AI chip family for three years in that time, but it's the first time I hear of chips "wearing under high intensity use".

Chips absolutely do wear out. Dopants electromigrate and higher temperatures make it faster. Discrete components like capacitors also tend to fail over time.

Is it going to be that significant though? No idea.

Re: After the AI boom: what might we be left with?

#124

Post by trenchpilgrim »

pclmulqdq wrote: post #122
trenchpilgrim wrote: post #88

Earlier quoted context omitted.

They're mostly solid state parts. The parts that do wear out like fans are easily replaced by hobbyists.

This is not correct. Solid-state parts wear out like mechanical parts, especially when you run them hot. The mechanism of this wear-out comes from things like electromigration of materials and high-energy electrons literally knocking atoms out of place.

Those parts take decades to wear out, excepting a manufacturing defect.

Re: After the AI boom: what might we be left with?

#125

Post by lossolo »

bitmasher9 wrote: post #56

> GPUs that have a 1-3 year lifespan In 10 years GPUs will have a lifespan for 5-7 years. The rate of improvement on this front has been slowing down faster then CPU.

I'm still waiting to get a used Nvidia A100 80 GB (released in 2020) for well under $10,000.

Re: After the AI boom: what might we be left with?

#127

Post by WalterSear »

rjh29 wrote: post #46
scellus wrote: post #17

He writes as if only datacenters and network equipment remain after the AI bubble bursts. Like there won't be any AI models anymore, nothing left after the big training runs and trillion-dollar R&D, and no inference served.

Who's going to pay to run those models? They are currently running at a huge loss.

Anthropic said their inference is cash positive. I would be very surprised if this isn't the norm.

Re: After the AI boom: what might we be left with?

#128

Post by trollbridge »

archerx wrote: post #16

I believe the next step will be robotics and getting A.I. to interact with the physical world at human fidelity. Maybe we can finally have a Rosie from the Jetsons.

Robots still haven’t come close to replicating human and animal touch as a sense, and LLMs don’t do anything to help with that.

Re: After the AI boom: what might we be left with?

#129

Post by muldvarp »

Juliate wrote: post #45
scellus wrote: post #17

He writes as if only datacenters and network equipment remain after the AI bubble bursts. Like there won't be any AI models anymore, nothing left after the big training runs and trillion-dollar R&D, and no inference served.

The point is, after the bubble burst, will there be enough funds, cash flow and... a viable market, to make these still run?

Inference is not that expensive. I'd argue that most models are already useful enough that people will pay to run them.

Re: After the AI boom: what might we be left with?

#130

Post by ogogmad »

dotnet00 wrote: post #120
mikert89 wrote: post #107

Earlier quoted context omitted.

the difference is the impact of crypto was always hypothetical, chatgpt can be used, explored, and if you are creative enough, levered in ways as the ultimate tool

You've done nothing but reuse the Sam Altman/Elon Musk playbook of making wild and extremely vague statements. Maybe say something concrete? What's a positive real world impact of LLMs where they aren't hideously expensive and error prone to the point of near uselessness? Something that isn't just the equivalent of a crypto-bro saying that their system for semi-regulated speculation (totally not a rugpull!) will end…

So you're saying that modern LLMs are a just like crypto/Web3, except in all the ways they're not, so they must be useless.

---

Less flippantly, they are excellent for self-studying university-level topics. It's like being able to ask questions to a personal tutor/professor.

Post reply on HN