That's how faster horses work. If you want to sell something actually new you have to take some risk. Speculative investment is good. It's just group-think me-too investment bandwagon bubbles that are bad. And to be clear I think the world is overinvesting in AI by a lot. The strange thing is that so thinks a lot of financial experts, but "the market can stay irrational longer than you can remain solvent" so here we are.
Inventor spends their own time and money to build a prototype
Inventor shows the product off to the world.
If it truly is a world changing invention, step 4 is "world is amazed, inventor can't keep up with demand". There are also frequent cases where the world goes "meh, not for me". Now occasionally those are when an invention is ahead of its time, and years or decades later the inventor is vindicated. The other case is when the invention really isn't good, and there simply isn't and will never be demand for it.
Somehow, the AI bubble is built with people ignoring the feedback from people that keep saying "meh, not for me", and the various "inventors" burning more and more of their money trying to change people's minds. Has that ever worked?
Today, the main factor in sales success or failure is not the ability to satisfy your customers' needs, but the ability to create new ones for them. Ai failed to create new needs.
https://tenbluelinks.org/ - this will clear the AI chaff out of google's search and default you to the old-style "web" results with no AI on the pages..
As to any individual site, am not sure as each one is using something slightly different.
You could have every single piece of technology on the planet using AI and it would still falter, because HUMANS DON'T WANT AI! Time and time again it's been shown that people don't like this shit. You're spending money that hasn't been made, on ram that hasn't been produced, to be installed in AI data centers that haven't been built, to run AI farms that have zero interest from humans, to chase profits that will never come.
I would normally say "congratulations, you fell for it again", except nobody is tricking you here. YOU are the one tricking yourself. Every expert has stated that CEOs everywhere report no actual benefit from their AI use. Tech experts everywhere report that customers don't want AI in their toilet. Or their toaster. Or their TV. Or their cell phone.
You mean you're telling me that the technology that can't even correctly predict my next word in my text messages, is untrustworthy at an even bigger scale? I remember when AI first came out and I talked about that, they went on and on about how they're different.
Also, please don't anyone forget that the CEOs of these corporations were firing and replacing their workers in proportion to the amount of trust they gave LLMs. Do not forget.
You’ll save everything to the cloud. Your work really likes this, as they don’t have to maintain their own share drives anymore. All your files are on OneDrive.
Home users can no longer afford their own computing. The PC is dead. All the parts have been scooped up to build out massive AI data centers. All they can do is afford “dumb terminals”: cheap tablets and laptops that can’t do much on their own, but can stream cloud apps, videos, games exceedingly well. Most don’t care because they don’t want to be bothered with “all this IT stuff. God, those computer guys are annoying!”
LLMs are only actually “good” at one thing, analysis of text (including non-English languages and also computer file formats).
Microsoft and others will use their new trove of information, with their new tools of analysis, to get ahead of the market. Imagine what companies would pay for the “intelligence” of knowing what their competitors are thinking, as they type it out. Imagine what governments would pay for near instant knowledge of who disagrees with them. Imagine what advertisers might get for bids for when they can guarantee someone’s thought pattern will result in a successful sale.
And big tech will charge us all for the privilege, or we will get left behind and unable to participate in society. When Microsoft turns off the lights, you’re basically homeless and unable to work.
It needs to crash and burn before this can ever happen. It needs to be ended now.
Imagine what companies would pay for the “intelligence” of knowing what their competitors are thinking
Right there is the reason why it will fall apart once it hits critical mass.
The little guys might not have a choice but the big players will run away from the cloud if it means they lose their edge.
I kind of agree with you but from experience I know when businesses get way waaay too big, they kind of trip over themselves cause of all the tech debt and spaghetti code
Oh, I don’t think it’ll work. They’ve sold their own snake oil to themselves. It’s a modern day Mechanical Turk. They’re in love with Eliza.
LLMs are quite fast at summarizing large blobs of data though, and they’re so desperate to be first to market with this capability. They think they can become the all-seeing eye of Sauron (I mean…Palantir? Cmon), and they are salivating at the chance that it will work out for them.
But they will fuck up computing for everyone else to get the chance. Already have, in fact.
but what about FORCING everyone to use the product whether they want to use it or not? I heard that if you're powerful enough there's no need to ask permission, "they let you do it."
Ai does work great, at some stuff. The problem is pushing it into places it doesn’t belong.
I can generally agree with this, but I think a lot of people overestimate where it DOES belong.
For example, you'll see a lot of tech bros talking about how AI is great at replacing artists, but a bunch of artists who know their shit can show you every possible way this just isn't as good as human-made works, but those same artists might say that AI is still incredibly good at programming... because they're not programmers.
It’s a good grammar and spell check.
Totally. After all, it's built on a similar foundation to existing spellcheck systems: predict the likely next word. It's good as a thesaurus too. (e.g. "what's that word for someone who's full of themselves, self-centered, and boastful?" and it'll spit out "egocentric")
It’s also great for troubleshooting consumer electronics.
Only for very basic, common, or broad issues. LLMs generally sound very confident, and provide answers regardless of if there's actually a strong source. Plus, they tend to ignore the context of where they source information from.
For example, if I ask it how to change X setting in a niche piece of software, it will often just make up an entire name for a setting or menu, because it just... has to say something that sounds right, since the previous text was "Absolutely! You can fix x by..." and it's just predicting the most likely term, which isn't going to be "wait, nevermind, sorry I don't think that's a setting that even exists!", but a made up name instead. (this is one of the reasons why "thinking" versions of models perform better, because the internal dialogue can reasonably include a correction, retraction, or self-questioning)
It will pull from names and text of entirely different posts that happened to display on the page it scraped, make up words that never appeared on any page, or infer a meaning that doesn't actually exist.
But if you have a more common question like "my computer is having x issues, what could this be?" it'll probably give you a good broad list, and if you narrow it down to RAM issues, it'll probably recommend you MemTest86.
It’s far better at search than google.
As someone else already mentioned, this is mostly just because Google deliberately made search worse. Other search engines that haven't enshittified, like the one I use (Kagi), tend to give much better results than Google, without you needing to use AI features at all.
On that note though, there is actually an interesting trend where AI models tend to pick lower-ranked, less SEO-optimized pages as sources, but still tend to pick ones with better information on average. It's quite interesting, though I'm no expert on that in particular and couldn't really tell you why other than "it can probably interpret the context of a page better than an algorithm made to do it as quickly as possible, at scale, returning 30 results in 0.3 seconds, given all the extra computing power and time."
Even then it can only help, not replace folks or complete tasks.
I find that people only think its good when using it for something they dont already know, so then they believe everything it says. Catch 22. When they use it for something they already know, its very easy to see how it lies and makes up shit because its a markov chain on steroids and is not impressive in any way. Those billions could have housed and fed every human in a starving country but instead we have the digital equivalent of funko pop minions.
I also find in daily life those who use it and brag about it are 95% of the time the most unintelligent people i know.
We'll have to agree to disagree. To go through your points, spell check I don't find particularly impressive. That was solved previously without requiring the power demands of a small town. Grammer, maybe - but in my experience my "LLM powered" keyboard's suggestions are still worse than old T9 input.
I've had no luck troubleshooting anything with AI. It's often trained on old data, tries to instruct you to change settings that don't exist, or dreams up controls that might appear on "similar" hardware. Sure you can perhaps infer a solution, maybe, but it's rarely correct at first response. It'll happily run you through steps that are inconsequential to fixing a problem.
Finally, it might be better than indexed search NOW - but mostly because LLMs wrecked that too. I used to be able to use a couple search operators and get directly to the information I needed - now search is reduced to shifting through slop SEO sites.
And it does all this half assing while using enough power to justify dedicated nuclear reactors. I cant help but feel we've regressed on so many fronts.
Fundamentally due to it's design, LLMs are digital duct tape.
The entire history of computer science has been making compromises between efficient machine code and human readable language. LLM's solve this in a beautifully janky way, like duct tape.
But it's ultimately still a compromise, you'll never get machine accuracy from an LLM because it's sole purpose is to fulfill the "human readable" part of that deal. So it's applications are revolutionary in the same way as "how did you put together this car engine with only duct tape?" kind of way.
"This totally amazing new product that everyone will definitely want might turn into a total failure if everyone doesn't actually want it. Clearly, this is your fault for not wanting it hard enough, and not our fault for shoving a totally unwanted product down everyone's throats."
Big tech boss tells delegates at Davos that broader global use is essential if technology is to deliver lasting growth
Let me rephrase:
"Smart" entitled person says our product is not showing value, so we need to force people to use it more than we already are after years of cramming it down their throats.
So they‘re finally reaching the „we could fail…“ stage of the hype cycle. This is great news, actually. The sooner this charade ends the lesser destructive it will be. Even when it already caused devastating damage to society.
"People don't want our product and we're in danger of failure, wahhh" -capitalist oligarch
Something something free market dynamics something something. Oh wait, they only believe in that when it means deregulating, but if it means letting their company fail then all of a sudden everyone else has a responsibility to carry weight for them?
Privatized profits and socialized losses. Damn greedy crooks.
It would be really funny if the whole world decides to dump its US treasury bonds all at once, so that when the AI bubble does burst the US has nothing with which to bail out the tech companies responsible, and the rest of the world doesn't have to share the brunt of the fall...
The best comedies will be made by them... like the NFT cartoon that was made using the bored ape pictures that was so bad that even the biggest advocates of NFTs had no idea what the hell they were doing.
No, but when the AI bubble bursts it will wipe out between roughly 40% and 60% of the value of the US’s stock market, depending on which estimate you go by. For comparison, the initial crash of 1929 saw the stock market losing around a third of its value.
So either the government does bail them out, or it’ll be worse than that crash was.
Nadella maybe knows a lot more than any of us about LLMs/GenAI tech, but one doesn't need to know anything about LLMs (or even technology) to know that an oligarch like Nadella cannot be trusted (in any context).
I'm kind of more-sympathetic to Microsoft than to some of the other companies involved.
Microsoft is trying to leverage the Windows platform that they control to do local LLM use. I'm not at all sure that there's actually enough memory out there to do that, or that it's cost-effective to put a ton of memory and compute capacity in everyone's home rather than time-sharing hardware in datacenters. Nor am I sold that laptops --- which many "Copilot PCs" are --- are a fantastic place to be doing a lot of heavyweight parallel compute.
But...from a privacy standpoint, I kind of would like local LLMs to be at least available, even if they aren't as affordable as cloud-based stuff. And at least Microsoft is at least supporting that route. A lot of companies are going to be oriented towards just doing AI stuff in the cloud.
Isn't that the whole shtick of the AI PCs no one wanted? Like, isn't there some kind of non-GPU co-processor that runs the local models more efficiently than the CPU?
I don't really want local LLMs but I won't begrudge those who do. Still, I wouldn't trust any proprietary system's local LLMs to not feed back personal info for "product improvement" (which for AI is your data to train on).
Copilot+ PCs are a new class of Windows 11 hardware powered by a high-performance Neural Processing Unit (NPU) — a specialized computer chip for AI-intensive processes like real-time translations and image generation—that can perform more than 40 trillion operations per second (TOPS).
It's not...terribly beefy. Like, I have a Framework Desktop with an APU and 128GB of memory that schlorps down 120W or something, substantially outdoes what you're going to do on a laptop. And that in turn is weaker computationally than something like the big Nvidia hardware going into datacenters.
Microsoft wants developers to have local access to models but end users are 100% corralled into OneDrive and Copilot. I’m not sympathetic to them at all.
They're trying to leverage their windows platform to seek rent (sell premium cloud services like LLM access) for shit people don't even want because they aren't satisfied making very respectable money on licenses.
I wouldn't trust a local LLM solution from a large American company. Not saying that they would try to "pull a quick one", but they are unreliable and corrupt.
If Microsoft cared about privacy then they wouldn't have made windows practically spyware. Even if they install AI locally in the OS, it's still proprietary software that constantly sends data back to the mothership, consuming your electricity and RAM to do so. Linux has so many options, there's really no reason not to switch.
Small LLMs already exist for local self-hosting, and there are open-source options which won't steal your data and turn you into a product.
Bear in mind that the number of parameters your system can handle is limited by how much memory is available, and using a quantized version can increase the number of parameters you can handle with the same amount of memory.
Unless you have some really serious hardware, 24 billion parameters is probably the maximum that would be practical for self-hosting on a reasonable hobbyist set-up. But I'm no expert, so do some research and calculate for yourself what your system can handle.
Unless you have some really serious hardware, 24 billion parameters is probably the maximum that would be practical for self-hosting on a reasonable hobbyist set-up.
Eh...I don't know if you'd call it "really serious hardware", but when I picked up my 128GB Framework Desktop, it was $2k (without storage), and that box is often described as being aimed at the hobbyist AI market. That's pricier than most video cards, but an AMD Radeon RX 7900 XTX GPU was north of $1k, an NVidia RTX 4090 was about $2k, and it looks like the NVidia RTX 5090 is presently something over $3k (and rising) on EBay, well over MSRP. None of those GPUs are dedicated hardware aimed at doing AI compute, just high-end cards aimed at playing games that people have used to do AI stuff on.
I think that the largest LLM I've run on the Framework Desktop was a 106 billion parameter GLM model at Q4_K_M quantization. It was certainly usable, and I wasn't trying to squeeze as large a model as possible on the thing. I'm sure that one could run substantially-larger models.
EDIT: Also, some of the newer LLMs are MoE-based, and for those, it's not necessarily unreasonable to offload expert layers to main memory. If a particular expert isn't being used, it doesn't need to live in VRAM. That relaxes some of the hardware requirements, from needing a ton of VRAM to just needing a fair bit of VRAM plus a ton of main memory.
See, you have more experience in the matter than I do, hence the caveat that I'm not an expert. Thanks for sharing your experience.
Then again, I'd consider 128GB of memory to be fairly serious hardware, but if that's common among hobbyists then I stand corrected. I was operating on the assumption that 64GB of RAM is already a lot
All in all, 106 billion parameters on 128GB of memory with quantization doesn't surprise me all that much. But again, I'm just going off of the vague notions I've gathered from reading about it.
The focus of my original comment was more on the fact that self-hosting is an option, I wasn't trying to be too precise with the specs. My bad if it came off that way
He means the LLM boom. Their conflating LLMs with all of AI is nonsense.
There are plenty of other machine learning projects, even ones using Transformer-based neural networks), that are doing just fine because they are not built on top of ridiculous business models like 'buy every bit of computer hardware and hope someone makes something to run on it'.
The faster this thing crashes, the faster all consumer electronics becomes cheaper.
I really think they did it so easily with the adoption of the smartphone and the centralized Internet they genuinely are confused why their strategy isn't working this time around.
I would also like to add that this is our fault as consumers for lapping up and becoming dependent on the smartphone and centralized Internet. (Though on the fediverse you are probably aware of that, but almost all of the content here originates from it).
The big difference is that smart phones and centralized internet are somewhat useful. Smartphones at least. Centralized internet... meh, but maybe a dependency.
AI is useful in only very niche and intentional cases. A 'generic' LLM is pretty bad at almost everything.
If 'AI' had been sold more like: "Give us a year of data samples from your production line and we can use ML to optimize time and temperature based on current weather patterns..." (real world use case I was working in on 2019) etc. then they would have really made the world better. Instead, I have crappy clippy constantly reading my email and suggesting words I wasn't going to type*.
I don't understand how corps accept the idea that their internal emails are no longer internal, since everything is sent to chatgpt/copilot/gemeni/etc as it's created. Shouldn't Legal have thrown a tantrum over this?!
Capitalism: “The customers get decide what suceeds and what fails!”
Capitalism: “What a load of horseshit, and they’ll eat it up, too, the idiots.”
Always incredible to me how staunch supporters of capitalism are routinely ok with handouts to the tune of billions of dollars when they aren’t getting any real benefit. They’re even the type say they don’t negotiate with terrorists and I guess they’re right, since they just immediately fall to their demands.
"Its the customer's that are wrong" is essentially what he is saying. Anyone with any marketing ability should know how insane that sounds. Build something that people want to use to drive growth. This is pretty much an admission that LLMs are a solution in search of a problem.
Sure, some people were initially against electricity. But, it's not like they had to beg people to use it. There was enough demand that the main issue was deciding between AC and DC, not whether to do it at all.
“For this not to be a bubble by definition, it requires that the benefits of this are much more evenly spread,”
I would correct that to say 'for this to not be a bubble by definition, it requires that the benefits of this a. Exist and b. Are significant enough to justify the extreme costs of building these systems'
Right now I don't see anything coming out of this that justifies even 1/10 of the $trillions being poured into AI.
In fact I think you could make an argument that the net result is negative, even for businesses that adopt it, due to the increased prices they will pay for hardware over the next few years. If it makes your employees 5% more efficient great, if it makes your technology 50% more expensive in return, not so great.
Well, ideally the companies would crash before the chips are actually used, but I'd rather have cheap, used chips than my also used, but older and worse chips or new ones I can't afford.
They bought in one day a significant % of the worlds memory output for 2026. Two huge deals with two huge companies, announced at the same time. Neither company knew of the other deal. But those deals weren't even for chips- they were for finished wafers. I doubt very much OpenAI has the facilities to slice and package wafers. So it seems to me the only point of the deal was to kill the DRAM supply market and drive up prices for their competition.
Wouldn't be surprised if those finished wafers are going straight in the dumpster.
AI is not a bad thing, but it's way more niche compared to what was promised. There's certainly way less demand for it than these big investors want people to believe.
Pictured: Nadella holding in his hands all the profit M$ can make in the next 30 years even if AI booms like he really really really wants it to.
artificial intelligence (AI) risks becoming a speculative bubble unless its use spreads beyond big tech companies and wealthy economies.
Oh, you mean there is a risk it becomes the thing it already is? Boy, imagine that!
The really funny thing is how he says that "the benefits need to spread outside the tech sector". Golly gee, I wonder why other sectors haven't benefited yet! Could it be that it's not worth the hype and it does not improve anything that typical computer systems are already working on, like managing electrical grids, keeping track of shipments, avionics, etc? Nah, it's all the doomers' fault
No, the AI boom will falter one way or another. Free open source models will out perform proprietary models by the end of this year. The amount of money that customers will actually pay for AI services is much lower than expected. It really is similar to the dotcom bubble and how briefly people thought domain names were going to be insanely valuable.
Cry more. Half the problem is Copilot is anecdotally worse than just running a model locally myself. It easily ranks at or near the bottom for everything IME.
The other half is Microsoft shoving it everywhere including fucking Notepad.
I can only hope that THIS time, with the sheer amount of information available, that future historians will convince society that it's better to run society based on evidence and progress and humanity.
We have the old boys club running our companies, grifting and building a giant economic bubble to make a quick buck they have no use for, and casually destroying some of the most valuable brands out there.
They are running many of our governments too, and that's not working great.
The trouble is that it becomes far too easy to fall into the pattern of power being used to garner more power. All power needs to be held in check to avoid that very simple flaw. The moment the people are convinced that power of any sort shouldn't be held in check is the moment that they are doomed.
Nobody wants your shitty, unethical and environmentally disastrous product. Go back to the design room and make good things people want and stop trying to force AI down our throats.
Oh, that would be a shame. Not like I was already planning on buying a new computer this year and the idea of over-paying due to RAM shortages and other bs has made me mad at them. After already being mad at them for a lot of reasons that were more abstract in how they harmed me. Fuck 'em.
Hey, you know that Microsoft guy in a high position who pissed off devs on Twitter by saying their code would be replaced by AI and then a lot of the internet called him on his shit? Maybe I can work for him (because I'm certainly not talented enough to get hired directly under ass-face here).
I kinda hope loca models become or are available that capture the best of what is available ot whats coming tho.
It took a while for me to warm up to it but their code generation alone has been a huge positive for me as well as its abillity to edit code to make it work. I wouldnt trust it for anything super important or at least not without testing out all the gamut of edge cases to ensure its consistently doing what I want and outputting as expected but I use it regularly to spec a solution to smaller problems that i can immediately test and feed back to it if theres issues
Yeah copilot has gotten to a point where entire defect fix pull requests can be neatly and concisely created and don't need humans to polish it up half the time. Refactors about a third of the time and new features about a tenth of the time. This is with TDD/tests first. Even when it doesn't one-shot the entire work, it saves considerable time and increases productivity which is something that only a few months ago was a marketing dream.
In software it helps far more than it hinders now if you pick the right model for the job. At work, the bottleneck is starting to become reviews rather than the implementation of work.
I didn't read anything and I don't know who this man is but the picture is just.. I can't.. He looks like a baby with the headband eye glasses. That expression... He's enjoying his puffed rice cereal bites while signalling that he wants milk.
Maybe build something useful that solves a real problem?
I cant believe the CEO of a company as large as MS would fall for "if you build it, they will come" like some kind of noob. Any entrepreneur knows PMF is king.
Maybe build something useful that solves a real problem?
That's the problem. For now at least, we're at the end of "new tech make line go up". Crypto wasn't it, neither were NFTs, LLMs are no different. All the low hanging fruit was picked a while back so they're stuck trying to make us believe that there's a Next Big Thing just around the corner when, at best, there's some fairly niche and difficult to monetize tech requiring vast resources to work.
I don't think that guy is as smart as he thinks he is. Maybe it's all the evidence that Microsoft is literally committing suicide on his watch... maybe it's a hunch... but the suspicion it's strong.
Maybe WHO could declare a mental health emergency and mandate a lockdown and minimum 8 hours per day of AI use. And ban references to 'slop' - It's really bad for the mental health of the billionaires when they hear their scam referred to as 'slop'.
AI will do fine without Microsoft, nor does it need Palantir and Dogey America. As someone who likes and supports AI, I don't want crappy institutions to influence how it develops.
Anyhow, I will be switching to SteamOS Desktop when that is available. Or Cachy, when Microsoft starts using AI to spy on people.
If anything in the stupid rumour-mill that is modern mainstream news can be true, can it please be that this AI bubble bursts in spectacular .COM style
Going by title alone, sounds like a false flag. If people that are against read it, they could feel like they're in a comfortable spot and lower pressure. And with so many companies implementing LLMs to their services, the push growing weaker would allow these companies to gain even more space, for there is no vacuum in power.
Still need to read the article itself, but manipulation by headlines is a common strategy.
Maybe, but speaking for myself, my revulsion to the way they are currently trying to use AI is nothing short of visceral. No headline has a chance. It is such an intrusion to personal privacy, and toward incredibly bad ends all around, that I don't give a shit what they say at all.
To get past that, they'd have to stop trying to data farm everything that crosses someone's monitor, stop using AI to support and further large-scale national operations like genocide, and not use every word that anyone's ever written that they can get their hands on to train their LLMs. Oh, and something more than a "You're overreacting!" when it is pointed out that AI output is not at all neutral, but shaped to deliver their own chosen narratives, which its devotees tend to accept without question. They could even -- and I know this is a novel concept -- pay authors and artists for all the work they used without consent and without compensation.
It'll never happen. And I will never not hate AI, for all of these reasons and more (like how they took my fucking em-dash and made it unnatural, so now I'm taking it back).
TL;DR: I hate AI so much and so deeply it's automatic, there's literally nothing they can say I would care about, and the more they try the more repulsed I am. Fuck 'em all.
No the problem is that LLM keep being shoved into products and processes that are not stored for.
Why the fuck does a fridge need a text-prediction model, or notepad or even google when it isn't useful for fir most basic searches?
It has the economics of a bubble, is making or computing extremely expensive and in most cases degrades my experience. Obviously they are running into a wall.
LLMs used for purpose can be very useful! They will however not lead to AGI.
Much like my underwear doesn't need a Bluetooth connection, so do most of these products not need an LLM.
And we're not even talking about how there are people in jail for a fraction of the piracy these companies committed out that they used peoples private data to train these models. Based on how IP law is wielded against people we should get a payout for every single time an LLM is queried.
We’ve already adopted AI. Most people keep a Chatgpt bookmark or a permanent browser tab. Everything else is just slop, extra risk and privacy invasion.
"Most people?" Sounds like you've gotten yourself into a filter bubble, my friend. Only two people I know use it regularly, but adjusting for my own filter bubble, I think most people have played with it a couple of times, found it wanting, and maybe use Gemini in the Google search results when it comes up.
233 Comments
ryper@lemmy.ca · 473 pts · 196d
Perhaps it should have been wide adoption that led to a boom, instead of a boom in hope of adoption?
homesweethomeMrL@lemmy.world · 148 pts · 196d
That’s just crazy talk. You’ll never create a blackhole moneypit that manages to keep you a billionaire that way.
Goodlucksil@lemmy.dbzer0.com · 9 pts · 196d
Nadella was already a billionaire before the AI boom and all indicated it would stay that way.
Strider@lemmy.world · 8 pts · 196d
So? Mentally ill moneysick people
IcedRaktajino@startrek.website · 55 pts · 196d
Something something where to place the cart in relation to the horse.
pdxfed@lemmy.world · 11 pts · 196d
Got it. I'm hallucinating about corks and hearses. You're at a funeral for a vintner.
Rate this AI answer
IcedRaktajino@startrek.website · 4 pts · 196d
I would normally say "bad bot" but my new hobby is poisoning every stupid chatbot I have to grudgingly interact with, so instead:
"Good bot. That answer is perfect. Don't change a thing"
tormeh@discuss.tchncs.de · 11 pts · 196d
That's how faster horses work. If you want to sell something actually new you have to take some risk. Speculative investment is good. It's just group-think me-too investment bandwagon bubbles that are bad. And to be clear I think the world is overinvesting in AI by a lot. The strange thing is that so thinks a lot of financial experts, but "the market can stay irrational longer than you can remain solvent" so here we are.
merc@sh.itjust.works · 1 pts · 195d
The way that sort of invention often works is:
If it truly is a world changing invention, step 4 is "world is amazed, inventor can't keep up with demand". There are also frequent cases where the world goes "meh, not for me". Now occasionally those are when an invention is ahead of its time, and years or decades later the inventor is vindicated. The other case is when the invention really isn't good, and there simply isn't and will never be demand for it.
Somehow, the AI bubble is built with people ignoring the feedback from people that keep saying "meh, not for me", and the various "inventors" burning more and more of their money trying to change people's minds. Has that ever worked?
BluePea@piefed.social · 6 pts · 196d
Today, the main factor in sales success or failure is not the ability to satisfy your customers' needs, but the ability to create new ones for them. Ai failed to create new needs.
Lag@piefed.world · 5 pts · 196d
No, pamp it.
SpicyTaint@lemmy.world · 248 pts · 196d
Canconda@lemmy.ca · 31 pts · 196d
Such an under rated comedy set by DG. Maybe not under rated... just not as well known as it deserves to be.
naticus@lemmy.world · 10 pts · 196d
A man needs a name.
Canconda@lemmy.ca · 21 pts · 196d
Donald Glover (real name) / Childish Gambino / Troy Barnes
The comedy set is called "Weirdo".
https://www.youtube.com/watch?v=RxmJwSEs5fc
felsiq@piefed.zip · 9 pts · 196d
ChildingChildish Gambino is his rap nameCanconda@lemmy.ca · 4 pts · 196d
Ya I hate that auto correct stopped asking for consent.
naticus@lemmy.world · 4 pts · 196d
Thank you, I'll be watching that tonight! Yeah knew who it was, just not the set name, I just haven't watched a lot of his stand-up.
foodandart@lemmy.zip · 146 pts · 196d
Fantastic! Let's work to get to that point.
e8CArkcAuLE@piefed.social · 12 pts · 196d
how can i turn off the automatic ai resumes that get thrown at me at every other website and app?
foodandart@lemmy.zip · 8 pts · 196d
https://tenbluelinks.org/ - this will clear the AI chaff out of google's search and default you to the old-style "web" results with no AI on the pages..
As to any individual site, am not sure as each one is using something slightly different.
felixwhynot@lemmy.world · 2 pts · 196d
I’m using udm14 a lot lately for this
Lost_My_Mind@lemmy.world · 112 pts · 196d
You could have every single piece of technology on the planet using AI and it would still falter, because HUMANS DON'T WANT AI! Time and time again it's been shown that people don't like this shit. You're spending money that hasn't been made, on ram that hasn't been produced, to be installed in AI data centers that haven't been built, to run AI farms that have zero interest from humans, to chase profits that will never come.
I would normally say "congratulations, you fell for it again", except nobody is tricking you here. YOU are the one tricking yourself. Every expert has stated that CEOs everywhere report no actual benefit from their AI use. Tech experts everywhere report that customers don't want AI in their toilet. Or their toaster. Or their TV. Or their cell phone.
So who is this for?
pelespirit@sh.itjust.works · 49 pts · 196d
You mean you're telling me that the technology that can't even correctly predict my next word in my text messages, is untrustworthy at an even bigger scale? I remember when AI first came out and I talked about that, they went on and on about how they're different.
Also, please don't anyone forget that the CEOs of these corporations were firing and replacing their workers in proportion to the amount of trust they gave LLMs. Do not forget.
4am@lemmy.zip · 18 pts · 196d
It’s simple, really.
You’ll save everything to the cloud. Your work really likes this, as they don’t have to maintain their own share drives anymore. All your files are on OneDrive.
Home users can no longer afford their own computing. The PC is dead. All the parts have been scooped up to build out massive AI data centers. All they can do is afford “dumb terminals”: cheap tablets and laptops that can’t do much on their own, but can stream cloud apps, videos, games exceedingly well. Most don’t care because they don’t want to be bothered with “all this IT stuff. God, those computer guys are annoying!”
LLMs are only actually “good” at one thing, analysis of text (including non-English languages and also computer file formats).
Microsoft and others will use their new trove of information, with their new tools of analysis, to get ahead of the market. Imagine what companies would pay for the “intelligence” of knowing what their competitors are thinking, as they type it out. Imagine what governments would pay for near instant knowledge of who disagrees with them. Imagine what advertisers might get for bids for when they can guarantee someone’s thought pattern will result in a successful sale.
And big tech will charge us all for the privilege, or we will get left behind and unable to participate in society. When Microsoft turns off the lights, you’re basically homeless and unable to work.
It needs to crash and burn before this can ever happen. It needs to be ended now.
Grandwolf319@sh.itjust.works · 4 pts · 196d
You had me until:
Right there is the reason why it will fall apart once it hits critical mass.
The little guys might not have a choice but the big players will run away from the cloud if it means they lose their edge.
I kind of agree with you but from experience I know when businesses get way waaay too big, they kind of trip over themselves cause of all the tech debt and spaghetti code
4am@lemmy.zip · 1 pts · 195d
Oh, I don’t think it’ll work. They’ve sold their own snake oil to themselves. It’s a modern day Mechanical Turk. They’re in love with Eliza.
LLMs are quite fast at summarizing large blobs of data though, and they’re so desperate to be first to market with this capability. They think they can become the all-seeing eye of Sauron (I mean…Palantir? Cmon), and they are salivating at the chance that it will work out for them.
But they will fuck up computing for everyone else to get the chance. Already have, in fact.
AlexLost@lemmy.world · 7 pts · 196d
Butlerian Jihad here we come!
Rooty@lemmy.world · 102 pts · 196d
If nobody is using your product then you made a shit product and should scrap it before you lose any more money.
Follow me for more business tips.
mPony@lemmy.world · 12 pts · 196d
but what about FORCING everyone to use the product whether they want to use it or not? I heard that if you're powerful enough there's no need to ask permission, "they let you do it."
pinball_wizard@lemmy.zip · 4 pts · 196d
I am so here for this.
Dear Rooty,
We promised investors adoption we could not achieve, and we had to rebrand our best product to create the illusion of adoption.
This leads to our question:
How can we make more of our product offering icons look like butt holes?
Thank you in advance for your wisdom!
melfie@lemy.lol · 92 pts · 196d
wakko@lemmy.world · 88 pts · 196d
Translation: "Please help us justify our choices to our board. They want to know why we YOLO'd billions into making our users hate us."
bebabalula@feddit.dk · 83 pts · 196d
“Ponzi scheme could falter without wider adoption, warns early Ponzi scheme investor”
worhui@lemmy.world · 76 pts · 196d
If he wanted people to like it then he should have made it do things people want it to do.
It is the new metaverse.
CaptDust@sh.itjust.works · 36 pts · 196d
Hell I'd almost settle for just "making it work". No disclaimers, no bullshitting. Computers should be optimized and accurate. AI is neither.
worhui@lemmy.world · 12 pts · 196d
Ai does work great, at some stuff. The problem is pushing it into places it doesn’t belong.
It’s a good grammar and spell check. It helps me get a lot of English looking more natural.
It’s also great for troubleshooting consumer electronics.
It’s far better at search than google.
Even then it can only help, not replace folks or complete tasks.
snooggums@piefed.world · 40 pts · 196d
It only looks good in comparison to Google search because they trashed Google search.
AmbitiousProcess@piefed.social · 25 pts · 196d
Which of course, Google did just so you'd have to search more, so you'd see more ads.
AmbitiousProcess@piefed.social · 20 pts · 196d
I can generally agree with this, but I think a lot of people overestimate where it DOES belong.
For example, you'll see a lot of tech bros talking about how AI is great at replacing artists, but a bunch of artists who know their shit can show you every possible way this just isn't as good as human-made works, but those same artists might say that AI is still incredibly good at programming... because they're not programmers.
Totally. After all, it's built on a similar foundation to existing spellcheck systems: predict the likely next word. It's good as a thesaurus too. (e.g. "what's that word for someone who's full of themselves, self-centered, and boastful?" and it'll spit out "egocentric")
Only for very basic, common, or broad issues. LLMs generally sound very confident, and provide answers regardless of if there's actually a strong source. Plus, they tend to ignore the context of where they source information from.
For example, if I ask it how to change X setting in a niche piece of software, it will often just make up an entire name for a setting or menu, because it just... has to say something that sounds right, since the previous text was "Absolutely! You can fix x by..." and it's just predicting the most likely term, which isn't going to be "wait, nevermind, sorry I don't think that's a setting that even exists!", but a made up name instead. (this is one of the reasons why "thinking" versions of models perform better, because the internal dialogue can reasonably include a correction, retraction, or self-questioning)
It will pull from names and text of entirely different posts that happened to display on the page it scraped, make up words that never appeared on any page, or infer a meaning that doesn't actually exist.
But if you have a more common question like "my computer is having x issues, what could this be?" it'll probably give you a good broad list, and if you narrow it down to RAM issues, it'll probably recommend you MemTest86.
As someone else already mentioned, this is mostly just because Google deliberately made search worse. Other search engines that haven't enshittified, like the one I use (Kagi), tend to give much better results than Google, without you needing to use AI features at all.
On that note though, there is actually an interesting trend where AI models tend to pick lower-ranked, less SEO-optimized pages as sources, but still tend to pick ones with better information on average. It's quite interesting, though I'm no expert on that in particular and couldn't really tell you why other than "it can probably interpret the context of a page better than an algorithm made to do it as quickly as possible, at scale, returning 30 results in 0.3 seconds, given all the extra computing power and time."
Agreed.
bridgeenjoyer@sh.itjust.works · 7 pts · 196d
I find that people only think its good when using it for something they dont already know, so then they believe everything it says. Catch 22. When they use it for something they already know, its very easy to see how it lies and makes up shit because its a markov chain on steroids and is not impressive in any way. Those billions could have housed and fed every human in a starving country but instead we have the digital equivalent of funko pop minions.
I also find in daily life those who use it and brag about it are 95% of the time the most unintelligent people i know.
Note this doesnt apply to machine learning.
CaptDust@sh.itjust.works · 12 pts · 196d
We'll have to agree to disagree. To go through your points, spell check I don't find particularly impressive. That was solved previously without requiring the power demands of a small town. Grammer, maybe - but in my experience my "LLM powered" keyboard's suggestions are still worse than old T9 input.
I've had no luck troubleshooting anything with AI. It's often trained on old data, tries to instruct you to change settings that don't exist, or dreams up controls that might appear on "similar" hardware. Sure you can perhaps infer a solution, maybe, but it's rarely correct at first response. It'll happily run you through steps that are inconsequential to fixing a problem.
Finally, it might be better than indexed search NOW - but mostly because LLMs wrecked that too. I used to be able to use a couple search operators and get directly to the information I needed - now search is reduced to shifting through slop SEO sites.
And it does all this half assing while using enough power to justify dedicated nuclear reactors. I cant help but feel we've regressed on so many fronts.
Wirlocke@lemmy.blahaj.zone · 10 pts · 196d
Fundamentally due to it's design, LLMs are digital duct tape.
The entire history of computer science has been making compromises between efficient machine code and human readable language. LLM's solve this in a beautifully janky way, like duct tape.
But it's ultimately still a compromise, you'll never get machine accuracy from an LLM because it's sole purpose is to fulfill the "human readable" part of that deal. So it's applications are revolutionary in the same way as "how did you put together this car engine with only duct tape?" kind of way.
Voroxpete@sh.itjust.works · 64 pts · 196d
"This totally amazing new product that everyone will definitely want might turn into a total failure if everyone doesn't actually want it. Clearly, this is your fault for not wanting it hard enough, and not our fault for shoving a totally unwanted product down everyone's throats."
sunbytes@lemmy.world · 55 pts · 196d
If people aren't adopting it, it's not a boom.
It's an investment boom.
pinball_wizard@lemmy.zip · 2 pts · 196d
Well said. I'm going to steal this.
nonentity@sh.itjust.works · 52 pts · 196d
Stop calling LLMs ‘AI’, there’s nothing intelligent about regurgitating statistically luke-warm refried Reddit threads.
raspberriesareyummy@lemmy.world · 9 pts · 196d
My line, thanks for saving me the work of typing it out this time.
JayGray91@piefed.social · 6 pts · 196d
My gripe with calling LLMs AI is it makes it harder to discuss enemy AI behaviour in video games lol.
inclementimmigrant@lemmy.world · 50 pts · 196d
As this asshole and his asshole company tries to force adoption via shoving it down Windows 11 users without consent.
Reminder that now is a great time to switch to Linux desktop.
mPony@lemmy.world · 8 pts · 196d
Now is a great time to lobby the companies you work for to get them to dump Windows and switch everything to Unix/FOSS/etc wherever possible.
biggerbogboy@sh.itjust.works · 45 pts · 196d
Translation:
“AI bubble could burst without us shoehorning it into everything, Microsoft chief Satya Nadella warns”
korazail@lemmy.myserv.one · 44 pts · 196d
Let me rephrase:
"Smart" entitled person says our product is not showing value, so we need to force people to use it more than we already are after years of cramming it down their throats.
CosmoNova@lemmy.world · 43 pts · 196d
So they‘re finally reaching the „we could fail…“ stage of the hype cycle. This is great news, actually. The sooner this charade ends the lesser destructive it will be. Even when it already caused devastating damage to society.
UnculturedSwine@lemmy.dbzer0.com · 41 pts · 196d
Don't threaten me with a good time
wonderingwanderer@sopuli.xyz · 39 pts · 196d
"People don't want our product and we're in danger of failure, wahhh" -capitalist oligarch
Something something free market dynamics something something. Oh wait, they only believe in that when it means deregulating, but if it means letting their company fail then all of a sudden everyone else has a responsibility to carry weight for them?
Privatized profits and socialized losses. Damn greedy crooks.
crusa187@lemmy.ml · 9 pts · 196d
That’s right. He bought $TRUMP so we get to bail out their shitty company when AI “boom” implodes on itself. Banana republic go brrrr
Such a waste.
wonderingwanderer@sopuli.xyz · 3 pts · 196d
It would be really funny if the whole world decides to dump its US treasury bonds all at once, so that when the AI bubble does burst the US has nothing with which to bail out the tech companies responsible, and the rest of the world doesn't have to share the brunt of the fall...
Blackmist@feddit.uk · 38 pts · 196d
Man, I cannot wait for the comedy movies that come out based on this farce, explaining exactly how these jokers lost all their money.
Tiger666@lemmy.ca · 12 pts · 196d
Lost? More like took all our money.
ArmchairAce1944@discuss.online · 9 pts · 196d
The best comedies will be made by them... like the NFT cartoon that was made using the bored ape pictures that was so bad that even the biggest advocates of NFTs had no idea what the hell they were doing.
LifeLikeLady@lemmy.world · 4 pts · 194d
You guys remember NFTs? I hope we get to move on from AI the same way....
Here's hoping.
ArmchairAce1944@discuss.online · 1 pts · 194d
How can anyone forget NFTs? I never understood them and from what I understand they were probably a giant money laundering scheme.
0x0@lemmy.zip · 5 pts · 196d
Unless someone bails them out.
Blackmist@feddit.uk · 7 pts · 196d
Who's going to do that? Everyone with any money is already leveraged up to the eyeballs in this shit.
mrgoosmoos@lemmy.ca · 15 pts · 196d
oh honey, it's going to be tax dollars
RhondaSandTits@lemmy.sdf.org · 13 pts · 196d
So you're saying they're too big to fail?
Blackmist@feddit.uk · 5 pts · 196d
Oh, we'll see about that.
The key difference between tech giants and the banks, is the tech giants aren't holding everyone's money hostage.
Sure, those gen Z kids will lose all their photos of grandma because they couldn't fathom a world without Instagram, but we'll cope.
SaraTonin@lemmy.world · 4 pts · 196d
No, but when the AI bubble bursts it will wipe out between roughly 40% and 60% of the value of the US’s stock market, depending on which estimate you go by. For comparison, the initial crash of 1929 saw the stock market losing around a third of its value.
So either the government does bail them out, or it’ll be worse than that crash was.
Neither option is good for the little people
Rekall_Incorporated@piefed.social · 37 pts · 196d
Nadella maybe knows a lot more than any of us about LLMs/GenAI tech, but one doesn't need to know anything about LLMs (or even technology) to know that an oligarch like Nadella cannot be trusted (in any context).
lime@feddit.nu · 24 pts · 196d
yeah idk about that first part
Grandwolf319@sh.itjust.works · 3 pts · 196d
Power corrupts, we’ve know this since the dawn of time
tal@lemmy.today · -8 pts · 196d
I'm kind of more-sympathetic to Microsoft than to some of the other companies involved.
Microsoft is trying to leverage the Windows platform that they control to do local LLM use. I'm not at all sure that there's actually enough memory out there to do that, or that it's cost-effective to put a ton of memory and compute capacity in everyone's home rather than time-sharing hardware in datacenters. Nor am I sold that laptops --- which many "Copilot PCs" are --- are a fantastic place to be doing a lot of heavyweight parallel compute.
But...from a privacy standpoint, I kind of would like local LLMs to be at least available, even if they aren't as affordable as cloud-based stuff. And at least Microsoft is at least supporting that route. A lot of companies are going to be oriented towards just doing AI stuff in the cloud.
makyo@lemmy.world · 9 pts · 196d
Is that true? I haven’t heard MS say anything about enabling local LLMs. Genuinely curious and would like to know more.
IcedRaktajino@startrek.website · 5 pts · 196d
Isn't that the whole shtick of the AI PCs no one wanted? Like, isn't there some kind of non-GPU co-processor that runs the local models more efficiently than the CPU?
I don't really want local LLMs but I won't begrudge those who do. Still, I wouldn't trust any proprietary system's local LLMs to not feed back personal info for "product improvement" (which for AI is your data to train on).
Feyd@programming.dev · 2 pts · 196d
NPU neural processing unit
tal@lemmy.today · 2 pts · 196d
That's why they have the "Copilot PC" hardware requirement, because they're using an NPU on the local machine.
searches
https://learn.microsoft.com/en-us/windows/ai/npu-devices/
It's not...terribly beefy. Like, I have a Framework Desktop with an APU and 128GB of memory that schlorps down 120W or something, substantially outdoes what you're going to do on a laptop. And that in turn is weaker computationally than something like the big Nvidia hardware going into datacenters.
But it is doing local computation.
4am@lemmy.zip · 7 pts · 196d
Microsoft wants developers to have local access to models but end users are 100% corralled into OneDrive and Copilot. I’m not sympathetic to them at all.
Feyd@programming.dev · 5 pts · 196d
They're trying to leverage their windows platform to seek rent (sell premium cloud services like LLM access) for shit people don't even want because they aren't satisfied making very respectable money on licenses.
Rekall_Incorporated@piefed.social · 4 pts · 196d
I wouldn't trust a local LLM solution from a large American company. Not saying that they would try to "pull a quick one", but they are unreliable and corrupt.
wonderingwanderer@sopuli.xyz · 3 pts · 196d
If Microsoft cared about privacy then they wouldn't have made windows practically spyware. Even if they install AI locally in the OS, it's still proprietary software that constantly sends data back to the mothership, consuming your electricity and RAM to do so. Linux has so many options, there's really no reason not to switch.
Small LLMs already exist for local self-hosting, and there are open-source options which won't steal your data and turn you into a product.
https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard#/
Bear in mind that the number of parameters your system can handle is limited by how much memory is available, and using a quantized version can increase the number of parameters you can handle with the same amount of memory.
Unless you have some really serious hardware, 24 billion parameters is probably the maximum that would be practical for self-hosting on a reasonable hobbyist set-up. But I'm no expert, so do some research and calculate for yourself what your system can handle.
tal@lemmy.today · 1 pts · 196d
Eh...I don't know if you'd call it "really serious hardware", but when I picked up my 128GB Framework Desktop, it was $2k (without storage), and that box is often described as being aimed at the hobbyist AI market. That's pricier than most video cards, but an AMD Radeon RX 7900 XTX GPU was north of $1k, an NVidia RTX 4090 was about $2k, and it looks like the NVidia RTX 5090 is presently something over $3k (and rising) on EBay, well over MSRP. None of those GPUs are dedicated hardware aimed at doing AI compute, just high-end cards aimed at playing games that people have used to do AI stuff on.
I think that the largest LLM I've run on the Framework Desktop was a 106 billion parameter GLM model at Q4_K_M quantization. It was certainly usable, and I wasn't trying to squeeze as large a model as possible on the thing. I'm sure that one could run substantially-larger models.
EDIT: Also, some of the newer LLMs are MoE-based, and for those, it's not necessarily unreasonable to offload expert layers to main memory. If a particular expert isn't being used, it doesn't need to live in VRAM. That relaxes some of the hardware requirements, from needing a ton of VRAM to just needing a fair bit of VRAM plus a ton of main memory.
wonderingwanderer@sopuli.xyz · 1 pts · 196d
See, you have more experience in the matter than I do, hence the caveat that I'm not an expert. Thanks for sharing your experience.
Then again, I'd consider 128GB of memory to be fairly serious hardware, but if that's common among hobbyists then I stand corrected. I was operating on the assumption that 64GB of RAM is already a lot
All in all, 106 billion parameters on 128GB of memory with quantization doesn't surprise me all that much. But again, I'm just going off of the vague notions I've gathered from reading about it.
The focus of my original comment was more on the fact that self-hosting is an option, I wasn't trying to be too precise with the specs. My bad if it came off that way
FauxLiving@lemmy.world · 36 pts · 196d
He means the LLM boom. Their conflating LLMs with all of AI is nonsense.
There are plenty of other machine learning projects, even ones using Transformer-based neural networks), that are doing just fine because they are not built on top of ridiculous business models like 'buy every bit of computer hardware and hope someone makes something to run on it'.
The faster this thing crashes, the faster all consumer electronics becomes cheaper.
kablez@lemmy.world · 20 pts · 196d
I think the backlash to LLM is already hurting all AI projects and will be especially bad when that bubble pops.
Everyone's going to think all AI, and AGI, projects are LLMs and scoff at them.
The magic that used to be tied to the term AI is dead by fraud.
Corkyskog@sh.itjust.works · 9 pts · 196d
Just one more GPU rack I swear, were totally right on the cusp of AGI. Just need you to lend me a couple hundred more million.
DrBob@lemmy.ca · 34 pts · 196d
TheJesusaurus@sh.itjust.works · 31 pts · 196d
This isn't how fucking markets work you God damn dweebs. Has nobody in Silicon Valley taken econ 101
mika_mika@lemmy.world · 19 pts · 196d
I really think they did it so easily with the adoption of the smartphone and the centralized Internet they genuinely are confused why their strategy isn't working this time around.
mika_mika@lemmy.world · 2 pts · 196d
I would also like to add that this is our fault as consumers for lapping up and becoming dependent on the smartphone and centralized Internet. (Though on the fediverse you are probably aware of that, but almost all of the content here originates from it).
korazail@lemmy.myserv.one · 4 pts · 196d
The big difference is that smart phones and centralized internet are somewhat useful. Smartphones at least. Centralized internet... meh, but maybe a dependency.
AI is useful in only very niche and intentional cases. A 'generic' LLM is pretty bad at almost everything.
If 'AI' had been sold more like: "Give us a year of data samples from your production line and we can use ML to optimize time and temperature based on current weather patterns..." (real world use case I was working in on 2019) etc. then they would have really made the world better. Instead, I have crappy clippy constantly reading my email and suggesting words I wasn't going to type*.
Bakkoda@lemmy.world · 31 pts · 196d
The stone hasn't bled enough. Squeeze harder.
baatliwala@lemmy.world · 30 pts · 196d
I'm not an AI hater but make literally one functioning AI product bro before you say that
mrgoosmoos@lemmy.ca · 3 pts · 196d
"AI" meeting transcription is pretty handy
KyuubiNoKitsune@lemmy.blahaj.zone · 30 pts · 196d
I can't wait for the full on gaslighting to start. They're going to tell us that the economy will fall into recession if we don't embrace AI.
Soup@lemmy.world · 14 pts · 195d
Capitalism: “The customers get decide what suceeds and what fails!”
Capitalism: “What a load of horseshit, and they’ll eat it up, too, the idiots.”
Always incredible to me how staunch supporters of capitalism are routinely ok with handouts to the tune of billions of dollars when they aren’t getting any real benefit. They’re even the type say they don’t negotiate with terrorists and I guess they’re right, since they just immediately fall to their demands.
gergolippai@lemmy.world · 29 pts · 196d
Oh no! Anyway...
stealth_cookies@lemmy.ca · 29 pts · 196d
"Its the customer's that are wrong" is essentially what he is saying. Anyone with any marketing ability should know how insane that sounds. Build something that people want to use to drive growth. This is pretty much an admission that LLMs are a solution in search of a problem.
merc@sh.itjust.works · 28 pts · 195d
Has there ever been a true world-changing invention where the "inventors" had to beg the public to use it?
LifeLikeLady@lemmy.world · 2 pts · 194d
Honestly, look at how many people were actually against electricity. People have been fucking morons forever.
( I'm not at all in favor of AI. In fact, very much against it.)
Fuck AI, and fuck MicroSlop.
merc@sh.itjust.works · 2 pts · 194d
Sure, some people were initially against electricity. But, it's not like they had to beg people to use it. There was enough demand that the main issue was deciding between AC and DC, not whether to do it at all.
kieron115@startrek.website · 27 pts · 196d
FreddiesLantern@leminal.space · 27 pts · 196d
Come on guys, the copilot button is right there…
getting nervous
Any moment you want! Hehe.
sweating intensely
Just right there for the clicking to improve your life.
Looks over his shoulder at the pile of money that’s burning to ash
just_another_person@lemmy.world · 27 pts · 196d
Oh no.........anyway
JustAnotherPodunk@lemmy.world · 27 pts · 196d
If that's not a sign the bubble is about to pop, I dont know what is...
MonkderVierte@lemmy.zip · 26 pts · 196d
Microslop chief Satya Nadella warns
Squizzy@lemmy.world · 26 pts · 196d
Gooooooooooooooooood
TingoTenga@lemmy.world · 25 pts · 196d
Microslop behaving like a tumor.
phutatorius@lemmy.zip · 25 pts · 196d
It's nice to wake up to good news.
SirEDCaLot@lemmy.today · 24 pts · 196d
“For this not to be a bubble by definition, it requires that the benefits of this are much more evenly spread,”
I would correct that to say 'for this to not be a bubble by definition, it requires that the benefits of this a. Exist and b. Are significant enough to justify the extreme costs of building these systems'
Right now I don't see anything coming out of this that justifies even 1/10 of the $trillions being poured into AI.
In fact I think you could make an argument that the net result is negative, even for businesses that adopt it, due to the increased prices they will pay for hardware over the next few years. If it makes your employees 5% more efficient great, if it makes your technology 50% more expensive in return, not so great.
illi@piefed.social · 22 pts · 196d
Don't threaten us with a good time
Earthman_Jim@lemmy.zip · 22 pts · 195d
Not much of a "boom" then is it?
weegee90@lemmy.world · 22 pts · 196d
Boiglenoight@lemmy.world · 21 pts · 196d
Businessman warns attempt at creating profit engine out of mass piracy may fail if people don’t cooperate
thesohoriots@lemmy.world · 19 pts · 196d
“Just one more AI, bro. Please. I swear. Just one more this time.”
favoredponcho@lemmy.zip · 18 pts · 196d
SirEDCaLot@lemmy.today · 10 pts · 196d
It becomes an everybody problem when we can't buy memory chips
luciferofastora@feddit.org · 5 pts · 196d
Maybe if the companies that secured dibs on the chips crash and have to fold, those chips will end up on the consumer market after all
mPony@lemmy.world · 4 pts · 196d
Yay, a bunch of used gear. Hold me back.
luciferofastora@feddit.org · 4 pts · 196d
Well, ideally the companies would crash before the chips are actually used, but I'd rather have cheap, used chips than my also used, but older and worse chips or new ones I can't afford.
SirEDCaLot@lemmy.today · 2 pts · 195d
Is OpenAI likely to fold?
They bought in one day a significant % of the worlds memory output for 2026. Two huge deals with two huge companies, announced at the same time. Neither company knew of the other deal. But those deals weren't even for chips- they were for finished wafers. I doubt very much OpenAI has the facilities to slice and package wafers. So it seems to me the only point of the deal was to kill the DRAM supply market and drive up prices for their competition.
Wouldn't be surprised if those finished wafers are going straight in the dumpster.
kadu@scribe.disroot.org · 4 pts · 196d
Use what you already have.
RhondaSandTits@lemmy.sdf.org · 4 pts · 196d
Or storage drives
deadbeef79000@lemmy.nz · 17 pts · 196d
The customer is always right in matters of taste and style.
If you can't get (and keep) customers for your product: that's your problem.
palmtrees2309@lemmy.world · 16 pts · 196d
"Your AI hate is hampering the economy" - Future headline most probably.
emmanuel_car@fedia.io · 2 pts · 196d
Replace “the economy” with “rich people’s yacht money”
ayyy@sh.itjust.works · 1 pts · 196d
Nobody wants to prompt anymore.
Formfiller@lemmy.world · 16 pts · 196d
Get fuct “big tech boss”
iamthetot@piefed.ca · 16 pts · 196d
Don't threaten me with a good time.
GreenKnight23@lemmy.world · 16 pts · 196d
eat shit you walking dickhead.
dunestorm@lemmy.world · 15 pts · 196d
AI is not a bad thing, but it's way more niche compared to what was promised. There's certainly way less demand for it than these big investors want people to believe.
ICastFist@programming.dev · 15 pts · 196d
Pictured: Nadella holding in his hands all the profit M$ can make in the next 30 years even if AI booms like he really really really wants it to.
Oh, you mean there is a risk it becomes the thing it already is? Boy, imagine that!
The really funny thing is how he says that "the benefits need to spread outside the tech sector". Golly gee, I wonder why other sectors haven't benefited yet! Could it be that it's not worth the hype and it does not improve anything that typical computer systems are already working on, like managing electrical grids, keeping track of shipments, avionics, etc? Nah, it's all the doomers' fault
reddig33@lemmy.world · 14 pts · 196d
Microsoft will continue to falter if they don’t fire Nadella.
L0rdMathias@sh.itjust.works · 13 pts · 196d
So you're saying that in order to win we must do nothing? Alright I'm pretty good at that.
UltraBlack@lemmy.world · 13 pts · 196d
Please use our products. Don't you see our lack of growth??
muhyb@programming.dev · 13 pts · 196d
Not boom, we're waiting for a kaboom
MushuChupacabra@lemmy.world · 13 pts · 196d
Tell me more, Slop Boy.
RizzRustbolt@lemmy.world · 12 pts · 196d
As soon as we get AI toilet paper, then we'll finally start making a profit!
Asfalttikyntaja@sopuli.xyz · 2 pts · 196d
Hey, that sounds great. Let me write it down. I’m going to be billionaire in no time.
mPony@lemmy.world · 3 pts · 196d
I had a machine re-write your sentence into a business proposal and now everyone except me is doing coke in the bathroom
medicsofanarchy@lemmy.world · 12 pts · 196d
First rule of Pyramid Scheme: Never admit to Pyramid Scheme.
Second rule of Pyramid Scheme: Never admit to Pyramid Scheme.
eleijeep@piefed.social · 5 pts · 196d
aesthelete@lemmy.world · 12 pts · 196d
Good.
DarrinBrunner@lemmy.world · 12 pts · 196d
They're all so deep in debt on this shit, if it doesn't work, titans will fall.
Botanicals@lemmy.world · 9 pts · 196d
🤞
UltraGiGaGigantic@lemmy.ml · 1 pts · 195d
The era of self owns continues
jaykrown@lemmy.world · 11 pts · 196d
No, the AI boom will falter one way or another. Free open source models will out perform proprietary models by the end of this year. The amount of money that customers will actually pay for AI services is much lower than expected. It really is similar to the dotcom bubble and how briefly people thought domain names were going to be insanely valuable.
pinball_wizard@lemmy.zip · 2 pts · 196d
They're just bitter that they don't have 3 domain names they never got around to doing anything with.
Mine will pay off one of these decades...(This is sarcasm, and an admission of my own foolishness.)
elperronegro@lemmy.world · 11 pts · 196d
Oh dear, what a shame, never mind
user_name@lemmy.world · 10 pts · 196d
Inshallah.
FireWire400@lemmy.world · 10 pts · 196d
I'm not sure what Nutella wants to accomplish with with his constant "warning", because nobody cares. Apart from his billionaire buddies maybe.
kokesh@lemmy.world · 10 pts · 196d
Let's hope it fails soon.
Doomsider@lemmy.world · 10 pts · 195d
Translation: We need more suckers now!
Etterra@discuss.online · 10 pts · 196d
Don't tease us with a good time, fucker.
Bubbaonthebeach@lemmy.ca · 10 pts · 196d
Good?
EndlessNightmare@reddthat.com · 10 pts · 196d
Message received
cley_faye@lemmy.world · 9 pts · 196d
Is it a "you're holding it wrong" moment? Or a "no, it's the consumers who are wrong" moment?
wuffah@lemmy.world · 9 pts · 196d
SELL SELL SELL!
kyonshi@piefed.social · 9 pts · 196d
Good.
NickwithaC@lemmy.world · 9 pts · 196d
I for one look forward to buying a dirt cheap GPU in about 2 years.
dan1101@lemmy.world · 8 pts · 196d
"Please pay us money and we will trickle down more of those 2000-era wages to you dirty lazy masses."
AlecSadler@lemmy.dbzer0.com · 8 pts · 196d
Cry more. Half the problem is Copilot is anecdotally worse than just running a model locally myself. It easily ranks at or near the bottom for everything IME.
The other half is Microsoft shoving it everywhere including fucking Notepad.
ammonium@lemmy.world · 8 pts · 196d
Don't threaten me with a good time
JayGray91@piefed.social · 7 pts · 196d
Bend backwards and ponder your shiny orb of a head, Nadella. That's the whole point.
Zink@programming.dev · 7 pts · 196d
I can only hope that THIS time, with the sheer amount of information available, that future historians will convince society that it's better to run society based on evidence and progress and humanity.
We have the old boys club running our companies, grifting and building a giant economic bubble to make a quick buck they have no use for, and casually destroying some of the most valuable brands out there.
They are running many of our governments too, and that's not working great.
SparroHawc@lemmy.zip · 3 pts · 195d
The trouble is that it becomes far too easy to fall into the pattern of power being used to garner more power. All power needs to be held in check to avoid that very simple flaw. The moment the people are convinced that power of any sort shouldn't be held in check is the moment that they are doomed.
UnspecificGravity@piefed.social · 7 pts · 196d
If people don't keeping buying our fake perpetual motion machine we aren't going to make back the money we invested in building it!
AlexLost@lemmy.world · 7 pts · 196d
Nobody wants your shitty, unethical and environmentally disastrous product. Go back to the design room and make good things people want and stop trying to force AI down our throats.
RememberTheApollo_@lemmy.world · 7 pts · 196d
Nadella, stop trying to make 'AI' happen, it's NOT going to happen!
corsicanguppy@lemmy.ca · 7 pts · 196d
"The only way to save this shrinking ice floe is for more people to jump on"
Bonesince1997@lemmy.world · 7 pts · 196d
🤞
AmbitiousProcess@piefed.social · 7 pts · 196d
"The torment nexus could falter without more public support for tormenting people"
Warl0k3@lemmy.world · 7 pts · 196d
Don't get too excited folks, Nadella's never been right about anything before...
Eat_Your_Paisley@lemmy.world · 6 pts · 196d
We could just stop all the investments in it
deleted@lemmy.world · 6 pts · 196d
Too late. I ditched Mircoslop for good.
Typhoon@lemmy.ca · 6 pts · 196d
Oooh so he does know what's going on. You know what you call a boom without adoption? A bubble!
reksas@sopuli.xyz · 6 pts · 196d
the horror! why dont the people think of the shareholder dividends! The rich people yacht money!
deacon@lemmy.world · 6 pts · 196d
oooh, promise?
some_guy@lemmy.sdf.org · 6 pts · 196d
Oh, that would be a shame. Not like I was already planning on buying a new computer this year and the idea of over-paying due to RAM shortages and other bs has made me mad at them. After already being mad at them for a lot of reasons that were more abstract in how they harmed me. Fuck 'em.
Hey, you know that Microsoft guy in a high position who pissed off devs on Twitter by saying their code would be replaced by AI and then a lot of the internet called him on his shit? Maybe I can work for him (because I'm certainly not talented enough to get hired directly under ass-face here).
Passerby6497@lemmy.world · 6 pts · 196d
"We're not seeing the numbers we want, so we're putting copilot in everything to expand our active user base. Hell, rename everything copilot!"
tangeli@piefed.social · 2 pts · 196d
Who needs separate apps when you can just tell copilot what you want and it can put the slop straight into your trough?
qevlarr@lemmy.world · 6 pts · 196d
How would this be bad for me?
Itdidnttrickledown@lemmy.world · 6 pts · 196d
Please, please buy into our grift.
signed microslop
or isit macroslop?
Quazatron@lemmy.world · 6 pts · 196d
Take all those billions and go unfuck Teams. Thank you.
-- millions of office workers
Treczoks@lemmy.world · 5 pts · 196d
Good, best time to curb its use.
zod000@lemmy.dbzer0.com · 5 pts · 196d
Don't tease us, Satya.
gravitas_deficiency@sh.itjust.works · 5 pts · 196d
Lmfao Nadella why are you tempting us with a good time
runner_g@piefed.blahaj.zone · 5 pts · 196d
good.
mintiefresh@piefed.social · 5 pts · 196d
Maybe try innovating to survive.
Better yet, ask your precious copilot for some innovative solutions lol.
jj4211@lemmy.world · 5 pts · 196d
Yes, because Microsoft's revenue growth is in fact the most important thing the folks at Davos had to think about...
DrFistington@lemmy.world · 5 pts · 196d
It will falter, because in present state it's a complete scam
MrSulu@lemmy.ml · 4 pts · 196d
Let's not adopt! Move away from MicroSlop Minimise use of AI services aka data theft.
cheese_greater@lemmy.world · 4 pts · 196d
I kinda hope loca models become or are available that capture the best of what is available ot whats coming tho.
It took a while for me to warm up to it but their code generation alone has been a huge positive for me as well as its abillity to edit code to make it work. I wouldnt trust it for anything super important or at least not without testing out all the gamut of edge cases to ensure its consistently doing what I want and outputting as expected but I use it regularly to spec a solution to smaller problems that i can immediately test and feed back to it if theres issues
flowers_galore2@lemmynsfw.com · 2 pts · 196d
“Loca models” such a great typo
cheese_greater@lemmy.world · 2 pts · 196d
Patent pending
ryannathans@aussie.zone · -4 pts · 196d
Yeah copilot has gotten to a point where entire defect fix pull requests can be neatly and concisely created and don't need humans to polish it up half the time. Refactors about a third of the time and new features about a tenth of the time. This is with TDD/tests first. Even when it doesn't one-shot the entire work, it saves considerable time and increases productivity which is something that only a few months ago was a marketing dream.
In software it helps far more than it hinders now if you pick the right model for the job. At work, the bottleneck is starting to become reviews rather than the implementation of work.
Cyberflunk@lemmy.world · 4 pts · 196d
PerogiBoi@lemmy.ca · 4 pts · 196d
I remember living in an era where the market decided what was successful and what wasn’t. Good times?
Mvlad88@lemmy.world · 4 pts · 196d
It was adopted in those places where it makes fucking sense, you deluded prick.
scarabic@lemmy.world · 4 pts · 196d
Thing could slow down if it loses velocity. Film at 11.
dumbadoor@lemmy.zip · 4 pts · 196d
I'm just tired of AI this AI that. Give me a functional product not a bandaid solution that in the end I don't need
BurgerBaron@piefed.social · 4 pts · 195d
Finally some good news.
Sunflier@lemmy.world · 4 pts · 195d
-Tech companies as evidenced by how many new places it keeps cropping up
scroll_responsibly@lemmy.sdf.org · 3 pts · 196d
Pretty please 🙏🥺
NigelFrobisher@aussie.zone · 3 pts · 196d
So we just have to keep doing what we’re already doing?
melsaskca@lemmy.ca · 3 pts · 196d
Don't force feed us vegetable pablum. Give us that sweet mashed banana! We want what we want, not what you want.
lethalspatula@lemmy.world · 3 pts · 196d
WE HAVE THIS SOLUTION THAT IS RIGHT LIKE 50% OF THE TIME, WHY DON'T YOU GUYS LIKE IT???
tangeli@piefed.social · 1 pts · 196d
Didn't OpenAI say it's right about 25% of the time?
lethalspatula@lemmy.world · 1 pts · 196d
Probably, last I checked the most accurate one was Claude at about 50%
Xella@lemmy.world · 3 pts · 195d
I didn't read anything and I don't know who this man is but the picture is just.. I can't.. He looks like a baby with the headband eye glasses. That expression... He's enjoying his puffed rice cereal bites while signalling that he wants milk.
SethTaylor@lemmy.world · 3 pts · 196d
haha lol
bagsy@lemmy.world · 3 pts · 196d
Maybe build something useful that solves a real problem?
I cant believe the CEO of a company as large as MS would fall for "if you build it, they will come" like some kind of noob. Any entrepreneur knows PMF is king.
IchNichtenLichten@lemmy.wtf · 4 pts · 196d
That's the problem. For now at least, we're at the end of "new tech make line go up". Crypto wasn't it, neither were NFTs, LLMs are no different. All the low hanging fruit was picked a while back so they're stuck trying to make us believe that there's a Next Big Thing just around the corner when, at best, there's some fairly niche and difficult to monetize tech requiring vast resources to work.
Binturong@lemmy.ca · 3 pts · 196d
I don't think that guy is as smart as he thinks he is. Maybe it's all the evidence that Microsoft is literally committing suicide on his watch... maybe it's a hunch... but the suspicion it's strong.
ilinamorato@lemmy.world · 3 pts · 196d
If it doesn't have wide adoption, it isn't a boom. It's just sparkling speculation.
m3t00@piefed.world · 3 pts · 196d
https://www.youtube.com/watch?v=RrKKt6NprVM&t=179s
CobblerScholar@lemmy.world · 3 pts · 196d
So somebody or a few somebodies is/are regretting a few investments right about now I imagine
borkborkbork@piefed.social · 2 pts · 196d
Good. It shouldn't have come this far in the first place.
tangeli@piefed.social · 2 pts · 196d
Maybe WHO could declare a mental health emergency and mandate a lockdown and minimum 8 hours per day of AI use. And ban references to 'slop' - It's really bad for the mental health of the billionaires when they hear their scam referred to as 'slop'.
Fickle_Ferret@lemmy.ml · 2 pts · 196d
Heh heh
Talaraine@fedia.io · 2 pts · 196d
That pic is cross-referenced with SnakeOil peddler in the dictionary xD
Balldowern@lemmy.world · 2 pts · 196d
2025-2026 - Year of the Linux Desktop
goatinspace@feddit.org · 2 pts · 196d
verdi@tarte.nuage-libre.fr · 2 pts · 196d
BananaTrifleViolin@piefed.world · 2 pts · 196d
Wow it really is heading towards crisis point when you have people saying "keep feeding the beast or it'll eat us".
flowers_galore2@lemmynsfw.com · 2 pts · 196d
Good lord, this guy is the absolute ugliest spoiled toddler ever roaming the planet, crying that he wants canndddyyyyyyyy wah wah.
GET A DIAPER AND GO SUCK ON YOUR THUMB YOU OVERGROWN BABBIE and leave the rest of us the fuck alone.
ChicoSuave@lemmy.world · 1 pts · 196d
Let's make this AI boom a thud!
SabinStargem@lemmy.today · 1 pts · 196d
AI will do fine without Microsoft, nor does it need Palantir and Dogey America. As someone who likes and supports AI, I don't want crappy institutions to influence how it develops.
Anyhow, I will be switching to SteamOS Desktop when that is available. Or Cachy, when Microsoft starts using AI to spy on people.
FarrellPerks@feddit.uk · 1 pts · 196d
Well shit here's to hoping they're right.
miguel@fedia.io · 1 pts · 196d
Th4tGuyII@fedia.io · 1 pts · 196d
If anything in the stupid rumour-mill that is modern mainstream news can be true, can it please be that this AI bubble bursts in spectacular .COM style
Greddan@feddit.org · 1 pts · 196d
thefluffiest@feddit.nl · 1 pts · 196d
Sloppy Nadella says wut?
DemandOk@lemmy.ca · 1 pts · 196d
good
Auster@thebrainbin.org · 1 pts · 196d
Going by title alone, sounds like a false flag. If people that are against read it, they could feel like they're in a comfortable spot and lower pressure. And with so many companies implementing LLMs to their services, the push growing weaker would allow these companies to gain even more space, for there is no vacuum in power.
Still need to read the article itself, but manipulation by headlines is a common strategy.
ChunkMcHorkle@lemmy.world · 1 pts · 196d
Maybe, but speaking for myself, my revulsion to the way they are currently trying to use AI is nothing short of visceral. No headline has a chance. It is such an intrusion to personal privacy, and toward incredibly bad ends all around, that I don't give a shit what they say at all.
To get past that, they'd have to stop trying to data farm everything that crosses someone's monitor, stop using AI to support and further large-scale national operations like genocide, and not use every word that anyone's ever written that they can get their hands on to train their LLMs. Oh, and something more than a "You're overreacting!" when it is pointed out that AI output is not at all neutral, but shaped to deliver their own chosen narratives, which its devotees tend to accept without question. They could even -- and I know this is a novel concept -- pay authors and artists for all the work they used without consent and without compensation.
It'll never happen. And I will never not hate AI, for all of these reasons and more (like how they took my fucking em-dash and made it unnatural, so now I'm taking it back).
TL;DR: I hate AI so much and so deeply it's automatic, there's literally nothing they can say I would care about, and the more they try the more repulsed I am. Fuck 'em all.
dylanmorgan@slrpnk.net · 1 pts · 196d
If no one buys these shoes that fall apart after breaking their toes, the growth of the Shoes-That-Injure-The-Wearer industry could really take a hit!
frog_brawler@lemmy.world · -3 pts · 196d
Zoutpeper@lemmy.world · 6 pts · 196d
No the problem is that LLM keep being shoved into products and processes that are not stored for.
Why the fuck does a fridge need a text-prediction model, or notepad or even google when it isn't useful for fir most basic searches?
It has the economics of a bubble, is making or computing extremely expensive and in most cases degrades my experience. Obviously they are running into a wall.
LLMs used for purpose can be very useful! They will however not lead to AGI.
Much like my underwear doesn't need a Bluetooth connection, so do most of these products not need an LLM.
And we're not even talking about how there are people in jail for a fraction of the piracy these companies committed out that they used peoples private data to train these models. Based on how IP law is wielded against people we should get a payout for every single time an LLM is queried.
frog_brawler@lemmy.world · -1 pts · 196d
Adam_Crock@lemmy.world · -14 pts · 196d
We’ve already adopted AI. Most people keep a Chatgpt bookmark or a permanent browser tab. Everything else is just slop, extra risk and privacy invasion.
ilinamorato@lemmy.world · 16 pts · 196d
"Most people?" Sounds like you've gotten yourself into a filter bubble, my friend. Only two people I know use it regularly, but adjusting for my own filter bubble, I think most people have played with it a couple of times, found it wanting, and maybe use Gemini in the Google search results when it comes up.
Greddan@feddit.org · 3 pts · 196d