[blenderob]

I'm disappointed to see this kind of cringey, ChatGPT-generated nonsense-Latin being posted and upvoted on HN.

Yes, I know the author added a disclaimer:

> PS: I don’t know any Latin, ChatGPT did the translation for me, so there may be some mistakes.

But there aren't just "some mistakes". The Latin is essentially nonsense. The closest meaning I can draw from that phrase is:

"Protect wife, given things, truth."

That's barely coherent, let alone proper Latin. Is "usorem" even a word?

Of course, it's your blog and you're free to post what you like. But I'd hope the HN community would be more discerning. Imagine if someone posted an article with nonsense Fortran code generated by ChatGPT, adding a note like:

"PS: I don’t know Fortran, ChatGPT translated this Python to Fortran for me, so there may be some mistakes."

If the Fortran didn't even compile, would we still upvote it? I doubt it.

And really, why write nonsense Fortran when you could just write clear Python, Go, or Rust that the community very well understands? Likewise, why attempt fake Latin when you can just write plain English, which most people here understand?

[jraph]

I found that particular aspect cringe too, I wouldn't trust ChatGPT to be good at translating to Latin and wouldn't expect it to work remotely well for this. I also wish people refrained from using LLMs at all.

But apart from this aspect, I suspect the whole idea / approach pleases / feels good and that's what gets upvoted.

edit: the HN post is flagged now, so… xD