this post was submitted on 21 Feb 2026
120 points (96.2% liked)

Fuck AI

5989 readers
960 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
 

Not a ragebait post.

I started thinking why I hate AI and it's mostly:

  • It is pushed down my throat very hard for what it does;
  • The unauthorized use of content on the internet;
  • The worsening of the environmental crisis;
  • The content it generates is shit.

I am wondering do you have other arguments against it?

you are viewing a single comment's thread
view the rest of the comments
[–] Wfh@lemmy.zip 9 points 1 day ago* (last edited 1 day ago)

First the obvious reasons, I won't spend much time on them: it's an ecological catastrophe, it's an economical trainwreck, it's an ethical nightmare by being trained on stolen content, it's the most powerful propaganda machine ever created and it's made and massively pushed by ultracapitalist nazis.

With that out of the way, and already 90% of my hate lays there, let's talk about the dangers in my trade: software engineering. Like all trades, it takes training, time, practice, experience, a little bit of talent and a huge willingness to be constantly learning to get good enough to sell your wares. And most importantly, it requires us to understand what's already there and to think before attempting any implementation.

A good, experienced software engineer should produce code that's reliable, maintainable, efficient, and does exactly what it's supposed to do. LLMs, even specialized ones like Claude code, fail spectacularly at all four categories. They code like destructive baby engineers with delusions of grandeur with hacky, brute-forced half-solutions that address the immediate problem with no consideration for existing code, edge-cases and legibility. Every pull request changes so much code that it's completely unauditable. If you have enough experience and awareness, you can wrangle them into writing exactly what you want but then, what's the point? You saved the time it would have taken to write the code yourself, but you've spent much more time babysitting the LLM instead and spent a ludicrous amount of resources. Besides, WPM has never been a metric to distinguish the best engineers. It's useful to babysit the most junior engineers, because it's how you help them grow. It's a complete waste of time and money to babysit an LLM.

My new boss is a true GenAI believer. And I mean kool-aid, rapture, heaven-and-hell believer. He's convinced at his core that his god will produce features better and faster than any of us. He doesn't want to believe anymore how inefficient software engineering is at its core. Writing code is only a tiny part of it. You need to understand the needs of the users, specify the features, plan, write, test, fix, test again and fix again and test again and again and then deploy. Scale comes after deployment, when a feature hits potentially millions of users.

They keep talking about productivity, but none of the C-suites seem to understand that productivity is a consequence of good software engineering, not a prerequisite. Let's go back to good code: it's modular, reliable, maintainable, efficient and does what it's supposed to do. It means that it's easy to add and change features without breaking the rest. It's easy to find bugs and fast to fix them. It doesn't needlessly hog resources so it's cheaper and faster to run. This is true productivity, not churning slop at an accelerated rate in the hopes that some of them will work as intended.

And finally because I already spent too much time writing this, it makes us lazy. As I'm forced to use an agentic IDE and I'm monitored on my usage, I make it do the useless stuff. I write the important code, and I'm glad to delegate writing the mandatory technical documentation that nobody reads anyway or boilerplate code. Anything where being an experienced software engineer has zero added value. But I know most of my peers, most of them much younger and less experienced than me, do not have the same discipline and also delegate writing massive amounts of code. And it's dangerous because of how comfortable it is. I'm happy to delegate shit I can't be bothered to write, juniors are happy to delegate writing code they can't be bothered to understand. Claude keeps saying "OK, now I have a clear picture of what's happening/what I need to do". It cajoles us into believing it actually knows what it's doing. It pushes us to give it more and more of our work and to trust it. And it's by design. The more you feel comfortable being lazy, the more you tell it to do what you're paid for, the more you get addicted to it. You stop learning. You stop thinking. You just go to your integrated chat and ask it to solve the problem you're being paid to solve. One day will come when you won't be able to do anything without it because you've stopped honing your skills, you've become too lazy to do the boring stuff, you've become too used to have a machine answer your every needs instead of doing it yourself. They push so hard to make us addicted to it because it's their only way out of the forthcoming crash.