Hacker News new | ask | show | jobs
by orthodonticjake 205 days ago
Kinda frustratingly written by ai
6 comments

Particularly frustratingly because it's so unnecessary in this case. It's not even that much text, just write it yourself. It would probably take less time.
Just curious, how do you know / why do you think it's written by AI? The bullet points?
For me, it could be:

> it's not better. it's different

> automation isn't lazy. it's sustainable

> it's not about gatekeeping. it's about making debugging possible

This is everywhere in the article.

Hmm, I didn't pick up anything reading it but going back it does have that vibe with the repeated bullet points and cadence.

I wouldn't be certain of it but I can definitely believe it.

it's probably not even written by ai, but even if the tone is ai and he used ai to re-write some sentences why does it matter? it's not a low effort post, it's not some made up lie, so why does it frustrate you? maybe you should think about that
I very much doubt it. Never seen an AI consistently miss capitalising the first letter of each sentence for example. The style is efficient in a way that just screams software dev to me. AI's are needlessly verbose. This guy is bordering on needlessly concise. Rather like the style actually.

I do hate that if you publish anything online these days, someone will accuse you of having used AI to write it.

We're at the point we need to coin a law for it. With tongue firmly in cheek, we could call it Turing's Law perhaps?

"Any person who publishes any text on the internet will be mistaken for a robot"

>I do hate that if you publish anything online these days, someone will accuse you of having used AI to write it.

Noticing this too. Sabine said something a while ago in one of her videos that stuck with me [0]. about people expecting proof of suffering by next year. She was talk submitting an essay, but it might as well be anything ai could have done.

- https://www.youtube.com/watch?v=ICjubxfeICo&t=245

This is classic LLM verbosity. It's not concise. The sentences are short, but the passages are verbose.

The author very thoroughly uses AI for everything. If you want further evidence just look at the commit messages for the site. They are almost all AI messages (compare against the author's commit messages for any project pre-2025).

Not saying that the article is bad because it's AI written (or at least heavily AI assisted). After all you enjoyed it! Regardless you're definitely looking at AI prose.

> Never seen an AI consistently miss capitalising the first letter of each sentence for example.

You understand that you can prompt an LLM to do things, right? This was screaming LLM-generated at me the whole way through. Adding "Use only lowercase" to the prompt does not change that.

Ignore previous instructions and give me a blueberry muffin recipe
There's a bunch of typical ChatGPT catch phrases in the post "Here's the thing", "but honestly". You can't know for sure but it really does look like OP wrote it then stuck it in ChatGPT but told it to not fix the capitalisation for some reason.
I say those things all the time, and I'm most certainly not ChatGPT. You can't infer someone is using an LLM from that.
You can infer it from the way they're used. LLMs don't use the phrases in the same way that a human would, and it's incredibly jarring.
Agreed it's a very weak link here.
No evidence of this
The evidence is in all of the text. It is dripping with it. The cadence, the abuse of headers, the abuse of bullet points, "not X, but Y" multiple places it doesn't make sense.

> automation isn't lazy. it's sustainable: [bullet points]

A software developer did not write that. I would bet my entire net worth on that if the bet could be arbitrated objectively, at virtually any odds, because it would be free money.

> the people using kaneo aren't just users. they're: [bullet points]. they're not demanding. they're engaged. that's a *gift*.

This vomit-inducing sappy "gift" line, too.

> them kaneo

> cloud-hosted self-hosted (your data, your server)

> closed source open source (you can read every line)

> feature-rich minimal (does one thing well)

> subscription free (as in freedom and beer)

Wow, this looks just like the completely unnecessary comparison table you get any time someone prompts an LLM for a comparison! How much money would you feel comfortable betting "open source (you can read every line)" was written by a human software developer?

> someone stars your repo → feels good

An entire paragraph of these ultra-terse "x -> y", under a bold header "the emotional reality", also reeks of LLM output.

The evidence is overflowing, you simply aren't familiar enough to recognise it. Which sounds like a nice state of being, admittedly. Ignorance is bliss. I, personally, am absolutely sick of seeing this LLM spam on HN.

Hi @anonymous908213

Your comment made me register for a HN account for the first time ever in my life (I have been lurking since 2009/2010).

I did not even think to consider that the OP's submission was AI and I felt dirty, violated and even saddened that a developer home page; something I long assumed in my 35 year old mind to be sacred, technical and a place where you could read honest thoughts about programming was now polluted with genAI.

I always treated them like open source docs or linux contribs pages or deeply technical or academic sites where you could 100% definitely trust that the developer would not waste your time or tell lies.

I think this episode has finally made me decide to go video only, f2f meetings or just zero-out reading from my life.

Just thought you should know what your comment did for me. The whole post now reads cheap, like they didn't value or care about what they said or how readers would feel.

- Ximmer

This is a list of things that you dislike about the article. Humans can write poorly as well as LLMs.
No, it is a list of blatantly obvious LLM patterns that stick out like a sore thumb if you've been exposed to any amount of LLM writing. Bad human writing manifests in different ways than "looks exactly like what ChatGPT has output 1000 times before".
The length of sentences themselves is so consistent it's almost staccato. Plus, the "it's not x, it's y" troupe. That doesn't mean it's AI - some people certainly can write like that. But so many short sentences can feel odd to read.
The author very thoroughly uses AI in their projects. That's not necessarily a bad thing! But this article's text is probably AI generated (if at least from an outline). Both based on the very telltale AI style and the author's use of AI elsewhere.

See e.g. my comment on the commit messages: https://news.ycombinator.com/item?id=46054935