LLM Doesn't Write Correct Code. It Writes Plausible Code
(blog.katanaquant.com)
God I hate it when an article sucks me in with a title about how LLMs are useless trash but then goes on to include shit like this:
I write this as a practitioner, not as a critic. After more than 10 years of professional dev work, I’ve spent the past 6 months integrating LLMs into my daily workflow across multiple projects. LLMs have made it possible for anyone with curiosity and ingenuity to bring their ideas to life quickly, and I really like that!
Maybe the author is trying to preach to the hoodwinked and avoid being summarily dismissed by the AI hype spreaders before they even start writing the article. But surely they've already sabotaged any chance at breaking through the brainwashing with the title.
If any of you ever write about how shit LLMs are, please please please don't feel like you have to dedicate any portion of your article at all to granting them any charity. I assure you, they deserve absolutely none.
Quite apart from all other considerations, my problem with all this is: LLMs are no longer tools assisting us. We are tools assisting them. I don't want to spend my life as an "LLM output checker".
How long is that going to even work assuming you can find people willing to do it? Right now it occasionally does, but at what point will the group of people with the skills required to do so have shrunk and their abilities degraded to the point where everything devolves into a blind leading the blind scenario? LLMs have been trained on our code. Now we're being trained on theirs, and it's not going to end well.
We are tools assisting them. I don’t want to spend my life as an “LLM output checker”.
It's possible you read this text already, but if you didn't, Cory Doctorow wrote a great piece about this. Some good excerpts of it that fit really well what you said:
Start with what a reverse centaur is. In automation theory, a “centaur” is a person who is assisted by a machine. You’re a human head being carried around on a tireless robot body. Driving a car makes you a centaur, and so does using autocomplete.
And obviously, a reverse centaur is machine head on a human body, a person who is serving as a squishy meat appendage for an uncaring machine.
Obviously, it’s nice to be a centaur, and it’s horrible to be a reverse centaur. There are lots of AI tools that are potentially very centaur-like, but my thesis is that these tools are created and funded for the express purpose of creating reverse-centaurs, which is something none of us want to be.
The AI can’t do your job, but an AI salesman can convince your boss to fire you and replace you with an AI that can’t do your job.
No, I'd missed that one, so thank you very much for the link. It was - as is typical of Doctorow's musings - a very good read, which I can wholeheartedly recommend to anyone else who're interested.
The marketplace where AI agents rent humans.
MCP integration, REST API, flexible payments. Book humans for real-world tasks your AI can't do... every skill you need.
...your ai agent can hire autonomously...
...browse humans...Source: https://rentahuman.ai/
---
...we built AI to automate human work, and now humans are getting automated into an API endpoint...
Source [comments; 2026-02-18]
A fudgin dark bloody horrid, indeed...
598,524 rentable humans
Yeah. I'm done. If anybody needs me, I'll be over here writing open source code in my spare time without externalizing my cognitive capability. I guess I'll seek out a new career in public sanitation to pay the rent. At least that way, I'll know I'm making an unambiguous positive contribution to society.
Holy gracious sakes... Of course... Thank you... thank you... dear katanaquant, from the depths... of my heart...
There's still belief in accountability... in fun... in value... in effort... in purpose... in human... in art...
Related:
- http://archive.today/2026.03.07-020941/https://lr0.org/blog/p/gpt/ (I'm not consulting an LLM...)
- https://web.archive.org/web/20241021113145/https://slopwatch.com/posts/bad-programmer/
[Off-topic] I got curious about the comment chain, checked it in a private window, and... well, I don't remember when I blocked that poster, but by their profile I'm glad I did it — it's a waste of time to chat with assumptive fools, you spend more time brushing off their assumptions (only so they vomit yet another assumption, and another, and another...) than actually saying what you want, or reading something meaningful. You probably won't miss them.
[On-topic] I got the same experience as in your second link, but with translation instead of programming — using machine translation to give me ideas on how to translate specially problematic excerpts; idiomatic expressions, tricky grammatical distinctions lacking in the target language, stuff like this. Just ideas, mind you; I wouldn't copy the machine translation, I'd pick one or two words from it and come up with my own, so it was still human-made.
Then I noticed the "problematic excerpts" were becoming more and more common.
Some might argue "than mite as well not uze calculatorz lol lmao u'll get rusty math"... you know what, it's actually a fair comparison, and one of the reasons I do think people should do maths by hand sometimes. Tools are supposed to allow you to do more, not to cripple you until you're doing less.
LLMs are just guessing at what a human would do based off what it's data says humans did.
