From Goldman Sachs' "Gen AI: Too Much Spend, Too Little Benefit?" www.goldmansachs.com/images/migra... - ThreadSky

edzitron.com • 4 hours ago

From Goldman Sachs' "Gen AI: Too Much Spend, Too Little Benefit?"

https://www.goldmansachs.com/images/migrated/insights/pages/gs-research/gen-ai--too-much-spend%2C-too-little-benefit-/TOM_AI%202.0_ForRedaction.pdf

Comments

specialkman.bsky.social•4 hours ago

I work with a major OEM trying to use Palantir to reduce part numbers used and the amount of work checking the AI and correcting mistakes has far outweighed and "time savings" the AI was supposed to solve

andylee42.bsky.social•4 hours ago

It costs $(Infinity) and produces exactly nothing useful, while destroying the planet. Good work, tech dicks.

silphid.bsky.social•4 hours ago

Imagine the Adobe Cloud model, but with ANY knowledge-based task, a subscription to do any kind of work... forever

beto.betoworks.com•4 hours ago

Well for a trillion-dollar problem to solve it's not like there aren't any out there: global poverty, housing crisis, world hunger... picks are aplenty.

And all of them, things AI can't, and won't, ever solve. But hey let's keep throwing more money at it just because.

jamsamcam.bsky.social•4 hours ago

Counterpoint Computer started off a expensive and became cheaper

starlightwishes.bsky.social•4 hours ago

And this is what they want to force on us.

edzitron.com•4 hours ago

BTW if you want my newsletter about this report (which was from July 2024!) here it is. One of my biggest.
https://www.wheresyoured.at/pop-culture/

michaelddwyer.bsky.social•1 hour ago

📌

sam.robotsfightingdinosaurs.com•4 hours ago

This is such an important insight because this is a pretty amoral view -- it's pretty simple dollars and cents "does this product make sense to invest in" stuff which carries much more weight under capitalism. it's an ungodly expensive solution to problems that don't exist, setting aside all else

numbcat9.bsky.social•4 hours ago

May I gently suggest that the Masters of the Universe (as Goldman Sachs folks, among others, used to call themselves) aren’t the ones with an informed idea on AI’s actual future potential? Those guys just aren’t very smart.

AI in 2025 is at its “precocious toddler” stage. In 2030? Very different.

jillehughes.bsky.social•1 hour ago

Finally, some rational thinking.

potter4ever.bsky.social•4 hours ago

can they make it more cost effective if they follow the Deepseek model?

absolutereality.blog•4 hours ago

About 20 years ago, a common refrain was that the rich were good because they provided jobs. Nowadays, more people believe rich people are good because they are rich. However, worth pointing out that rich people are torpedoing their value proposition by pushing out jobs for AI.

rswestmore.land•4 hours ago

That's a smart guy

jbreckmckye.bsky.social•4 hours ago

Unfortunately I think asking questions like "is this ai stuff even useful" and "how do i make a profit from something that costs me money even when users pay" is a niche position amongst the executive class

frankenmizer.bsky.social•4 hours ago

Mr. Coello is on the main beam in his statement. Having been a part of the ML/AI community when the Google transformer paper dropped, I can attest it changed the game.
But when Altman started shopping funding for ~25 new nuclear reactor builds, scalabilty dirtnapped.
It's become a VC cashgrab.

frankenmizer.bsky.social•4 hours ago

s/Coello/Covello

kankato.bsky.social•4 hours ago

One of the more level headed columns I've read today. TY for sharing!

mrdobalina.bsky.social•3 hours ago

I mean he kinda nails that one

bigbangblitz.bsky.social•2 hours ago

This guy really gets it.

AI is ideal for trivial things like scanning and iterative tasks, it will never at any point in the future ever be suitable for a complex task. You could have a computer the size of the Earth and it will still be fallible, with the extra presumption of it being infallible.

eschaton411.bsky.social•4 hours ago

this technology only makes sense if it takes us somewhere beyond financial value, where there’s an infinite supply of AI labor so that the labor theory of value holds no sway

rondomane.bsky.social•4 hours ago

can the AI machines manage and regulate themselves? do they handle their own repairs? do they write their own software?

eschaton411.bsky.social•4 hours ago

that’s the vision, yes

rondomane.bsky.social•4 hours ago

That is never happening, and if it does you're never going to see it.

eschaton411.bsky.social•4 hours ago

it already is happening in some industrial applications

captainahab.bsky.social•4 hours ago

yes, it's an AGI cult

problem is that the AGI/singularity cult has made themselves systemically important to the economy chasing tech which make never exist

captainahab.bsky.social•4 hours ago

at least when we were doing this with MBS there were like actual houses with actual people living in them

eschaton411.bsky.social•4 hours ago

agreed, it’d be one thing to see agi as one sector among all the saas products, but due to huge cash injections in 08 and during covid, there’s too much money and too few investment vehicles. a lot of pressure for this tech to work or the whole economy goes flat

blazeward.bsky.social•4 hours ago

"Blockchain!!"

except those folks have gone quiet.

"Web 3.0!!!"

uhm.

jbreckmckye.bsky.social•4 hours ago

Notice as well how "AI" hype emerged exactly as soon as Web3 and crypto collapsed

Filling the same vacuum

blazeward.bsky.social•4 hours ago

"Gosh!"

demoargenti.bsky.social•4 hours ago

It seems the types of problems LLMs *can* solve, such as discovering how proteins fold, are so niche and specialized that this tech will never become profitable at scale. Even if it *was* ethical to use AI, or if it was a good idea to outsource thinking and creativity to "AI" models.

philcohen.bsky.social•4 hours ago

The same mix of losers and predators who invested in blockchain will do the same here, and in ten years we’ll all wonder how they fell for it.

Comments

Posting Rules

Reply