Is AI Facing a Trough of Disillusionment?
(spectrum.ieee.org)
from flango@lemmy.eco.br to technology@lemmy.world on 05 Sep 19:22
https://lemmy.eco.br/post/16304616
from flango@lemmy.eco.br to technology@lemmy.world on 05 Sep 19:22
https://lemmy.eco.br/post/16304616
Is AI’s hype cycle leading us into a trough of disillusionment? Dive into the reality behind GPT-5’s launch and its impact on the tech world.
A recent MIT report on AI in business found that 95 percent of all generative-AI deployments in business settings generated “zero return.”
threaded - newest
These guys are such a joke, but they are the ones laughing $.
We are the joke
Yep.
Time to go work a service job for peanuts and then play video games when I get home until I fall asleep.
I sure can’t wait until election season so I can vote for whoever my rulers chose for me.
The same guy who said he only needed a “few” more trillion dollars to make it all work.
Just like NFTs, this will be remembered as the biggest grift of our generation so far
It really feels like a “biggest grift so far” moment and they’re just desperate for a new hype to suck ungodly amounts of money into. Quantum computing, biotech or whatever they come up with.
I do think this will be the last traditional tech bubble. Not because VCs have learned any better, but because we’re at the limits of monetizing silicon under capitalism. The money will be dumped someplace else.
There are plenty of other software approaches that could happen, but capitalism doesn’t have a path to make money off of them.
I wouldn’t be surprised if the comments by Putin and Xi about immortality would trigger a rush on biotech and longevity projects.
True,I remember nfts died cause they lost their value alot.
Nfts never really had any use in the first place. Ai and LLMs does have a lot of fields where it will be used and expanded on, how much depends a lot on how it keeps developing.
But current state LLMs already can change a lot, but it will take years to be implemented widespread. Think of all the things that are still not digital at all
I'm guessing it'll be more like the dot com bubble in the 90s - too much too soon but it's not going away.
Exactly this. It’s the hype cycle all over again. Everyone gets excited about the new thing, the new thing is easy to implement but hard to utilize effectively, everyone starts shoehorning the new thing into their thing, they throw shit-tons of money at it and shit-tons more to market it as the greatest thing ever, and then everyone starts to see the truth. The new thing is okay and can be useful but it has been way overblown, and it all starts to crumble down to where it should be and then we build it up slowly from there.
i know it’s popular to be very dismissive, but a lot of “AI” has already been integrated into normal workflows. AI autocomplete in development text editors, software keyboards, and question asking bots isn’t going away. speech-to-text, “smart eraser”, subject classification, signal processing kernels like DLSS and frame generation, and so many more will be with us and improving for a long time. Transformers, machine learning optimized chips, and other ML fields are going to be with us for a long time. the comparison to NFTs is either angst or misunderstanding.
That’s the issue, AI right now means LLMs not deep learning/ML.
The Deep Learning/ML stuff will keep chugging along.
but LLMs do represent a significant technological leap forward. i also share the skepticism that we haven’t “cracked AGI” and that a lot of these products are dumb. i think another comment made a better analogy to the dotcom bubble.
ETA: i’ve been working in ML engineering since 2019, so i can sometimes forget most people didn’t even hear about this hype train until ChatGPT, but i assure you inference hardware and dumb products were picking up steam even then (Tesla FSD being a classic example).
We definitely haven’t cracked AGI, that’s without a doubt.
But yeah, LLMs are big (I’d say really Transformers were the breakthrough). My point though was that Deep Learning is the underlying technology driving all of this and we certainly haven’t run out of ideas in that space even if LLMs may be hitting a dead end.
I feel like LLMs didn’t hit a deadend so much as people started trying to use them in completely inappropriate applications.
I think it’s a mixture of that and the fact that when OpenAI saw that throwing more data drastically improved the models, they thought they would continue to see jumps like that.
However, we now know that bad benchmarks were misleading how steep the improvements were, and much like with autonomous vehicles, solving 90% of the problem is still a farcry away from 100%.
Nearly all of those can run just fine on-device. I think the part of the bubble that’s ripe to burst is the gigantic gigawatt data centers; we don’t even have the power to run them if all the ones under construction were completed. The current trajectory is not sustainable, and the more contact it has with reality the harder that will be to ignore.
NFTs weren’t that successful.
Sure, some morons got scammed. But AI is literally scamming governments.
Its only a trough if it goes back up on the other side.
Yeah, no one knows how to use it at scale in the enterprise without consultants setting it up for you.
I hope it’s more of a cliff. A trough implies that it will go back the other way.
Can’t close the Pandora’s box. AI is already out there and it’s being used in a variety of places. Some of those things are good, bad, ugly, useless, stupid or even terrifying. Once AI reaches the plateau of productivity, all the useless and stupid applications are gone, but the other ones will stay.
That all depends upon if they keep throwing dollars into the furnace to run the models.
“Useful” stuff will have to do an evaluation once they’re paying the true cost for these things as well, because most people are not running their own models…this shit is more centralized then even regular cloud computing.
I wasn’t talking about delusional LLMs and creepy image generators.
AI as a whole encompasses so much more. Like, image resolution up scaling, cleaning a noisy audio signal, generating new frames to make choppy stop motion animation smoother, finding optimal paths for logistics chains, designing a smart way to pack a pallet full of random boxes just to name a few. All of that stuff is already happening and is here to stay.
Cramming silly LLMs into every application might take a hit though. They could still be a solution to something, but we just haven’t quite figured it out yet.
Yeah, but they have tied AI to LLMs. If the investors that inflated this bubble sour on LLMs, the investments for AI more broadly speaking will probably dry up with it.
That’s true. The general sentiment of AI will influence investor behavior and the availability of money. If LLMs fail spectacularly, it could make it harder to find investors open to the idea of using AI for various other things too. The past 10 years has seen a massive rise in new AI companies, so they will be able keep on selling tried and tested AI even when investor money dries up.
Another AI winter? Another AI winter? (Possible in the medium or long term)
Although we’ll also have to wait and see what John Carmack does, because he’s also involved in this.
I think we may see an agentic AI winter, but there are so many other applications and systems that can benefit from deep learning still.
I hope the winter lasts till I die. Godfuck I want a peaceful life.
I remember when companies hung on to old technology for dear life as long as they possibly could. I'm pretty sure they were still making money then. Everyone hopping on these bandwagons and throwing tons of money at it for FOMO are bonkers, this stuff has barely been tested yet as far as productivity, effect on workers, mental health issues, the list goes on.. but I guess that's all fine.
We have an enormous part of the population retiring right now, some are voluntarily working right now and hanging on.
But these people will sell their stock en masse once they retire to buy something more stable like bonds.
The only way the stock market can keep them is to increase the yield so much that they get FOMO and don’t convert to bonds just yet.
This raises the stakes and winds up the stock market but now you have a HUGE portion of the population holding back a literal TSUNAMI of liquidation.
If the yield drops, if the apparent value slips, they will sell, the price will drop faster they will sell even more, this is how the next recession will happen, we just can’t say when.
When prices crash,the oligarchs will be there to gobble up everything of actual value.
In short AI valuations are holding back the next recession, and if they don’t deliver … it might be a depression and russification of the economy ?
They’re also crashing the stability of bonds
As a Russian, what’s that? If you mean oligarchy, you already have it. If you mean unseen parties holding power, you already have it. If you mean the down trajectory and deindustrialization, then perhaps a bit.
Your assesment is correct however the concentration of power has not yet reached the levels found in Russia, at least I don’t think so and as for the general readership they would disagree on whether this is really a problem at all yet, rather than just free market meritocratic competition
I think it’s pretty close, except there’s more pluralism of power.
I wonder if there would be, violent consolidations, forced mergers, corporation and regulators feeding on each other in a some kind of orgy of bureacratic excess
its so stupid how the companies seem just incapable of doing reasonable things. Either they hung on to old stuff even if it causes more problems or they jump on to some new hype thing. World would be better place if people with no long-term planning capability couldnt get any real influence over anything.
Now they are talking about replacing teachers wholesale with AI. We are getting closer to the end game and complete control of the human mind.
This makes me think we’re looking at the other side of the uncanny valley
This technology is equivalent to the atomic bomb for population manipulation. This is why the wealthy are so obsessed. They are getting closer to being able to control everyone by modeling our behavior.
The scariest part is no one is talking about the reality of this technology and what it will be used for. They have completely captured the narrative already.
To be fair, schools are already indoctrination mills.
I would not be concerned if it actually was a black hole of disillusionment…