July 16, 2026
Ctrl+Alt+Delusion
The LLM Critics Are Right. I Use LLMs Anyway
Even the loudest AI skeptics keep clicking send, and the comments are having a meltdown
TLDR: A writer says AI critics are right about the downsides, but people still use these tools anyway — and even AI builders are now blocking floods of low-quality submissions. Commenters are split between fear of brain-rot, worries about trust, and frustration that nobody can agree where useful help ends and harmful dependence begins.
This story hit a very online nerve: a writer says the critics of large language models — the text-generating AI tools behind chatbot coding assistants — are basically right about everything... and yet people still use them constantly. At a Berlin conference, the mood was peak contradiction. A respected software creator openly said his team auto-closes almost all outside submissions because they’re drowning in AI-made junk, while the crowd reportedly applauded anti-AI takes with AI tools still open on their laptops. The internet immediately recognized the vibe: not hypocrisy, exactly — more like a giant, collective "I know this is bad, but here we are."
And the comments? Absolutely feasting. One camp zeroed in on the fear that relying on AI will turn everyone’s brain into pudding, with one commenter comparing it to muscles that shrink when you stop using them. Another said the real issue isn’t whether AI touched the code at all, but whether anyone can still trust what they’re looking at. Then came the philosophical sniping: are “AI critics” even a real tribe, or has the internet invented cartoon enemies again? And the apprentice drama got especially spicy when one commenter dunked on the idea that seniors no longer need juniors, snapping that “master craftsmen didn’t take on apprentices to give them chores.” In other words: everyone agrees the mess is real, nobody agrees on what to do, and the comments section is serving full-blown group therapy with knives out.
Key Points
- •The author says they use LLMs frequently despite agreeing with many common criticisms of them.
- •At Local-First Conf in Berlin, the author observed both strong public criticism of LLMs and visible use of AI coding tools by attendees.
- •Armin Ronacher said Earendil auto-closes almost all pull requests and issues related to its Pi.dev project because of low-quality contribution volume.
- •The article highlights a trust problem in open source: maintainers can no longer easily infer meaningful human effort from a pull request.
- •The author lists several criticisms they consider valid, including copyright concerns, environmental costs, ethical issues, and financial bubble dynamics around AI companies.