this post was submitted on 17 May 2026
888 points (99.2% liked)
Fuck AI
7053 readers
1436 users here now
"We did it, Patrick! We made a technological breakthrough!"
A place for all those who loathe AI to discuss things, post articles, and ridicule the AI hype. Proud supporter of working people. And proud booer of SXSW 2024.
AI, in this case, refers to LLMs, GPT technology, and anything listed as "AI" meant to increase market valuations.
founded 2 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
LLMs absolutely have a market. It's a market that's probably in the low tens of billions of dollars worldwide. Non-critical translation, content writing, image generation, mockup. Etc etc. there are absolutely uses for LLMs and other generative AI.
The problem is that hundreds upon hundreds of billions have been invested, with hundreds more on the way. The current investment will never be recovered.
And these aren't durable investments, the equipment has a very limited shelve life, the chips aren't going to last nearly long enough to recover their cost, and the data centers are already obselete for the next generation.
Edit: getting downvotes for recognising that there is in fact a market for generative AI, by people who didn't catch that "the low tens of billions" is about 1% of what AI boosters are proclaiming.
Hard disagree on those two examples. That market is exclusively made up of people who a) can't write or make images themselves, and b) have an utter disregard for people who spent their lives doing so.
I do, however, agree completely with your points re the feasibility of returns on investments into "AI", and the expiration dates on the technical infrastructure. We have been presented with a solution without a preexisting problem — but oh boy, what exciting new problems comes after! /s
LLMs shouldn't be used in place of artists because LLMs can't make art. There's a quote I heard somewhere that sums this up perfectly: "LLMs should be used to help creative people do tedious things, not to help tedious people do creative things."
That's a very good way of framing it, yes! But at the moment it feels like the tedious people are at the wheel.
And that market is, unfortunately, absolutely huge. There is a frighteningly large demand for mediocre crap.
The LLM age has shown us how quickly the market is willing to settle for "not nothing" instead of "something good".
And a worldwide market in the 20 billion range is by no means remotely near the multiple trillions needed to actually fulfills what the AI boosters are promising.
You're on to something there. During the "AI" hype, I've seen several people claim that this could be the last gasp for late stage capitalism. Growth has plateaued, and they did not have anything more to sell us at a profit.
So they introduce a new artificial need, and promote it like their continued existence depended on it... because it does. But that was exactly my original point: nobody needs "AI", except the people peddling it.
In my experience, there are a lot of places very content to do a pretty bad job of stuff if it's low effort on the part of the person collecting the surplus labour value, and is making money.
The greatest deprogramming I've had through corporate experience is that capitalist myth that competition forces companies to try and make a better product.
Maybe in the odd case it does. But, for the most part, capitalism is pretty happy to just race to the bottom. A 2% worse product that costs 5% less to manufac6that I can sell for the original price? THAT is what every company is trying to do. Make a better product? Get the fuck outta here.
There is a market, for AI slop... because it's just a special case of a preexisting major market of "slop". We already had shovelware. Low effort high volume content creators. Telemarketers with accents so think you have to wonder if the costs saved by offshoring those workers could possibly recoup the reduced sales rates. It's all just slop. AI can absolutely disrupt the Slop Market. And it is.
But, as the other commenter pointed out... it's a tens-of-billions global industry, and AI is being invested in as if its a trillions global industry.
They need to make AI better, or make the slop industry larger. They won't meaningfully do the first, but can make some headroom on the second. Either way, not nearly enough for profitability.
Right now my money is on industry to repurpose barely used datacenter hardware which will become available for pennies on the dollar back to consumer-usable hardware.
Although I agree that there will always be produced "content" that is worse than other, I don't think there's any cause to compare human labour to "AI" slop. People can be stuck in jobs they don't care about, most often because of unreasonable work conditions or wages.
"AI" produces slop because that's what it is made for. That's its dire success criteria, making something that's statistically just above but not quite terrible quality. Manufacturers and employers prefer that over human labour performing similarly, i.e. quiet quitting. Because so far "AI" doesn't have workers' rights. Or higher aspirations.
And I cannot agree with your point about "telemarketers with accents" as a marker of low quality. That phrasing is a whole thicket of weeds that I'm not wading into.
I'm actually not saying anything is "better" than anything else. I'm saying there is a distinct industrial strategy that involves maximizing engagements primarily through volume of attempts at the minimal cost. It's very much a bimodal reality, clustered around trying to maximize completions against a constrained set of touches, or merely maximizing touches. I didn't mention it earlier, but "scams" also fit this profile. And I'm just calling it all slop. It's the cheapest thing that you can feed the pigs that'll eat anything anyways. Maximize channel saturation at the absolute lowest cost possible, the message itself being largely irrelevant.
AI excels at that specific task. You can agree or disagree all you want but studies trying to establish "is anyone actually seeing material profit gains out of AI", are seeing this exact pattern. People trying to use the slop machine to do anything besides making slop aren't having a great time. People who's business model is slop are doing great.
Unfortunately, those things are pretty much the only things it's good at and the companies losing billions on them are going to try to recoup as much as possible by doing as much of that as they can.
No. It most definitely is not.