this post was submitted on 23 Jul 2025
325 points (99.4% liked)

Fuck AI

3544 readers
390 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.

founded 1 year ago
MODERATORS
all 47 comments
sorted by: hot top controversial new old
[–] Vex_Detrause@lemmy.ca 58 points 3 days ago (3 children)

One in six US workers pretends to use 'cocaine' to please the bosses

[–] guyoverthere123@lemmy.dbzer0.com 14 points 3 days ago* (last edited 3 days ago)

pretends to use cocaine

[–] peoplebeproblems@midwest.social 14 points 3 days ago (1 children)

Yeah after reading that I can't some stop seeing it and it's hilarious.

[–] Nougat@fedia.io 7 points 3 days ago (2 children)

Someone needs to make a browser extension that changes " ai " to " cocaine " everywhere it appears.

[–] bridgeenjoyer@sh.itjust.works 5 points 3 days ago

Cloudtobutt 2.0

[–] peoplebeproblems@midwest.social 1 points 3 days ago (1 children)

Yep. If I had the slightest clue (or time) I'd start it.

[–] Pieisawesome@lemmy.dbzer0.com 2 points 3 days ago

You could use AI to build it 🤣

[–] burgerpocalyse@lemmy.world 4 points 3 days ago

i think its hitting. what a rush!

[–] Furbag@lemmy.world 20 points 3 days ago (2 children)

Why does it seem like the people who understand the least about AI and LLMs are the ones pushing it the hardest? Fuck the management class and fuck AI enshittification too.

[–] RunawayFixer@lemmy.world 12 points 3 days ago

They hope to make office staff redundant by replacing them with ai. The managers/consultants that are now pushing for LLM to replace office staff, are the same people that pushed for outsourcing office work to underpaid staff in low wage countries (and the companies that tried that, invariably got what they paid for). With both the resulting quality of work is far worse, but on paper it will save the company money. And if their customers are trapped in the short to middle term as is often the case with software companies, then the worse service won't even immediately affect the bottom line and the boss will have ample opportunity to jump ship to another cushy position before the bottom line does take a dive.

[–] Atropos@lemmy.world 7 points 3 days ago (2 children)

Be cause those of us who understand it, also understand how limited the actual uses are.

I think it has value as a fairly nice search engine. But not much else.

[–] zipzoopaboop@lemmynsfw.com 1 points 2 days ago

I've had good luck with zooms ai for taking notes. That's about it

[–] turtlesareneat@discuss.online -3 points 2 days ago* (last edited 2 days ago)

Spits out 500 line Powershell scripts, entire iOS apps in a single prompt... I realize what community this is but come on now

edit: You sound ignorant as hell when you can't even correctly define the thing you hate. But OK, have fun with that.

[–] ZDL@lazysoci.al 26 points 3 days ago (1 children)

My boss tried to get everybody to use LLMbeciles for work. He was taken in by the fakery and set up a commercial OpenAI account.

I refused to use it. Ever. Like to the point that I never even collected my ID and password.

Others used it, but it was, as always, a time-waster where it looked like they were making gains at the start, only to get bogged down when important details entered the picture. So I quietly got my work done while others wrestled with AI. Then by the end of six weeks nobody was using it and my boss quietly shut down the account.

[–] shalafi@lemmy.world 8 points 3 days ago (2 children)

Good for him! I'm wondering how many managers stick with AI due to sunk cost fallacy.

[–] ZDL@lazysoci.al 3 points 3 days ago

My boss is pretty decent. He gets dumb ideas (like we all do), but he recognizes when they fail and doesn't try to force them for face-saving. It's because of that that we survived the COVID-19 disasater and the pivot away from American clients (my initiative) without disappearing. He doesn't appear to have the usual ego of a businessman.

[–] zipzoopaboop@lemmynsfw.com 2 points 2 days ago

Or upward brown nosing

[–] supersquirrel@sopuli.xyz 57 points 4 days ago (1 children)

Is this the corporate more professional version of faking an orgasm?

[–] slaacaa@lemmy.world 9 points 3 days ago

“I’ll have what she’s having”

[–] AlecSadler@lemmy.blahaj.zone 25 points 3 days ago

I pretend to use it, but struggle. Like, I'll say it got me 80% of the way there but then I spend the rest of the day reviewing, fixing comments, adjusting unit tests...

...when secretly I was code complete in like 1-2 hours so I can just chill the rest of the day.

Management doesn't seem to question it because AI, and I'm pretty sure everyone else on my team is doing it too so we've sort of collectively created a low bar of expectations of output.

[–] peoplebeproblems@midwest.social 29 points 3 days ago (2 children)

Lmfao

That's me. LLMs slow me the fuck down. I have a theory though, LLMs only work for people who are able to not think through everything. The socially unacceptable feeling is really: I'm not dumb enough to benefit from LLMs.

But the reality is idk what the fuck LLMs can actually do

[–] okamiueru@lemmy.world 3 points 3 days ago* (last edited 2 days ago)

I'm impressed at the almost perfect correlation with the ones that consistently struggled with the fundamentals

[–] bridgeenjoyer@sh.itjust.works 2 points 3 days ago

I think you've cracked it. My mind overthinks everything in detail so automatically, that there's no way AI right now can keep up. Modern computers (with windows and work-spyware installed) are hardly fast enough for the amount of tasks I always have open at once. But I know a good amount of dumb people who treat the llm like a god and don't understand why they still get shit wrong, because they could never fully think for themselves in the first place.

This comment sounds assholeish but I am honestly just describing my day to day. Obligatory FuckAI.

[–] expatriado@lemmy.world 31 points 4 days ago (1 children)

at my job we suppose to know what we are doing, saying we use ai to solve problems would be more of a confession of incompetence than a brag

[–] bridgeenjoyer@sh.itjust.works 3 points 3 days ago

See, I have this exact way of thinking. Using a child's toy is not something to brag about yet they try to set it as a goal at work. Its so fucking backwards.

[–] Nougat@fedia.io 31 points 4 days ago (3 children)

I've used ChatGPT a couple of times to give me some powershell direction. Every time, it was wrong on kind of simple things. Like, "invented a cmdlet from whole cloth" kind of wrong.

[–] Lucidlethargy@sh.itjust.works 2 points 3 days ago

Yeah, they invent commands all the goddamn time. I fucking hate how much they lie.

[–] GrindingGears@lemmy.ca 3 points 3 days ago

So basically you "word-googled" how to do something.

[–] AlecSadler@lemmy.blahaj.zone 2 points 3 days ago (2 children)

ChatGPT is pretty awful for code in my experience. I actually disabled it in Cursor. Copilot is also garbage (because it is sorta just ChatGPT).

[–] TheEighthDoctor@lemmy.zip 2 points 3 days ago

I like cloud, it still creates imaginary methods and libraries but for very simple code or works decently

[–] GrindingGears@lemmy.ca 1 points 3 days ago

Copilot is like if you asked ChatGPT to create ChatGPT, but then pulled half the plugs, shorted out a bunch of wires and cut the budget by 90%.

I got a work email yesterday that was basically an advertisement for Gemini. Apparently my work will be holding contests using it. There are no details yet, but I find it all absurd. The work I do is in-person and client-facing, I don't know what they plan to even use an LLM for in this scenario. Maybe for writing notes? Our notes contain private, confidential medical information. No way in hell am I feeding that into Gemini.

I'm not shy about taking an ethical stand point (I left a job last year because upper management created an unethical environment.) I'm prepared to conscientiously object the forced use of AI, if it comes down to it. Between my clients' right to privacy, Google's lack of concern for privacy, and the horrific environmental impact that builds AI in the first place, this entire venture is highly questionable. I have to wonder how much money changed hands for this to even come about.

[–] dmention7@midwest.social 7 points 3 days ago (1 children)

I try to make it work for me--I really do. But it's just never worth the time & effort.

Recently, I asked our company-hosted instance of Copilot to draft up an application note for a new product we are releasing to help get my documentation juices flowing. This product is well documented on our internal sharepoint/teams/outlook/onedrive servers that Copilot scrapes, as well as some public facing press releases and marketing content, so I asked Copilot to draft up an app note for this product specifically by name.

Its first draft had tons and tons of info that was just straight-up wrong or super vague. So I refined the prompt asking it to cite sources for each paragraph. Lo and behold, there were 9 citations, and 8 of them were general technology references or competitor products.

Maybe it's a skill issue, but I can't get over the fact that this shit is not just so factually wrong, but so confidently wrong.

[–] PhilipTheBucket@quokk.au 3 points 3 days ago

Copilot is awful. Pretty much all the ones that operate by that model of "continuously sending pings to the server for various random requests and helpful hints" are pretty much guaranteed to be garbage, because the economics don't work unless they pare down the resources devoted to it until it's dumb as a post.

(Well, the economics still don't work even then, but they're hemorrhaging money at a little bit slower rate if they make it stupid)

Copying and pasting code to Claude, I've found to work decently well if you structure it so you can keep your focus and not put Claude in a position to fail too badly by taking on too large of a task. I know Anthropic has a code assistant that works on some kind of slightly different paradigm, so maybe that one is decent, I don't know. But yeah Copilot will never accomplish much of anything helpful beyond auto-completing some statements every now and then.

[–] misteloct@lemmy.dbzer0.com 15 points 3 days ago (1 children)

AI works really well for my domain (software engineering) especially for obscure one off languages, anywhere we're unfamiliar. Problem is I'm an overqualified senior and familiar with everything. I've sometimes fished for prompts that give a correct answer as required by my job lol, since I can verify it's correct. Turns out I'm also smart and fast at cheating the requirement.

[–] ThirdConsul@lemmy.ml 3 points 3 days ago (2 children)

I use it to write barely passable terraform.

God I hate hcl with passion.

[–] corsicanguppy@lemmy.ca 1 points 3 days ago

Hcl is hard, yeah.

But I'd work it all day vs fucking yaml. :-D

[–] jordanlund@lemmy.world 11 points 4 days ago (1 children)
[–] GrindingGears@lemmy.ca 5 points 3 days ago

13 out of 6 workers do this, only 1 in 6 is willing to admit it out loud.

[–] drosophila@lemmy.blahaj.zone 4 points 3 days ago* (last edited 3 days ago)

This is an example of how appearance-based and feeling-based a lot of the activities of businesses are.

If you look at previous social orders its very obvious how much activity served a social function vs a real physical need related to survival. The pyramids, for instance. While there's a bunch of things you could say about their role in the ancient Egyptian religion, and the effect of the make-work on Egypt's economy/society, I don't think anyone could argue that a giant pile of stones could physically help anyone put food in their belies or keep them warm at night; and when you get right down to it it seems pretty clear to me that the root cause of the pyramids is the ego of the guy in charge.

And yet there are many people who will tell you that everything a modern business does is maximally efficient. I would argue that this is pretty clearly not the case. One of the most blatant examples is dress codes and air conditioning. Cooling a large office building is not a small expense, and yet many businesses opt to have their employees wear hot suits (even their non-public facing employees) and turn down the temperature lower than it otherwise would need to be. You could extend that idea to the design of the building itself: a glass box is not the easiest thing to heat and cool. You could even extend it to the existence of the office building in general: while larger buildings are easier to heat and cool per unit area than a bunch of small ones (because of the square cube law), and there are certainly benefits to agglomeration, many office buildings are enormously tall and therefore enormously expensive (as construction costs do not grow linearly with height). Additionally, a lot of these buildings are built on some of the most expensive land that exists, which balloons their cost even more. I'm not saying that I think high rises are completely useless, but I do wonder if they need to be as common as they are, especially now that the internet exists (but lots of other people have talked about that).

Its interesting to note though that because businesses are supposed to be efficient, and the more efficient they are the better they are (more powerful, cunning, brutal, manly, etc) businesses adopt an aesthetic of efficiency. The use of LLMs is one example of this (using a new technology has the aesthetic of efficiency even if it measurably makes productivity worse) but it extends to a lot of the things businesses do. Even a lot of their architecture and industrial design fetishizes efficiency without actually being efficient, IMO.

I'm one of them

[–] bridgeenjoyer@sh.itjust.works 2 points 3 days ago

One of us one of us one of us

[–] doingthestuff@lemy.lol 3 points 3 days ago (1 children)

I can fake ai pretty well with my shitty photo editing skills.

[–] AnUnusualRelic@lemmy.world 1 points 3 days ago

I find that adding fingers isn't as easy as it seems.

[–] _lilith@lemmy.world 3 points 3 days ago

know enough about it to comfort the worried and placate the enthusiastic. Throw some buzzwords in there for them they love those