

I love it when someone so clearly projects.
You don’t like being corrected, do you?
Is it a shit translation? Yes. Is it “completely wrong”? No. Will you be able to accept something you said as incorrect, untrue? Also no. ;>
How many times did you go over your comment to try to make it as cryptic as you possibly can, when I just do it with ease? Must be annoying. Finns are always so fucking sensitive about someone being better than them at something. It’s actually built into Nordic culture, we just made it way worse.
It’s called the Law of Jante and it’s quite sad how little you realise it’s influencing you.
Perhaps expose yourself to more international culture? Learn a language or two. I did. After having learned all the dialects in Finland as a kid through different variations of Don Rosa stories and whatnot. (Ever read “Uutissii Turust”?)
Now you can of course make your text purposefully contextually missing to the LLM. It’s not hard, you’re kinda getting there. All you need is a lil practice, honey. And practice begins by understanding you’re not perfect. Ie, you can make mistakes. Like saying that “the translation was completely wrong”. It really wasn’t. It was a bad translation, and the first sentence was wrong, but with AI, you definitely get the gist that someone’s just using the language to communicate while trying to prevent others from understanding.
Too bad you have to write such long comments yet to ne able to confuse a simple LLM, but again, with practice, you’ll get there. ;>








Which said anything about it being a “working translation”? It renders context. You can infer from that much context, that someone is clearly using a dialect to obscure their message so as that the LLM would be confused.
Ah, I forget, Finns are just so bad at reading between the lines. (see Law of Jante from the other reply).
No, speaking in a dialect doesn’t actually make it impossible for an LLM to understand you. Just like you’ve seen when you rewrote your Finnish comment again and again and fixed this and that. Like I said, the Northern dialects aren’t as good in that as SW ones, because in the SW dialects, words are intentionally shortened, giving the LLM less data overall. And since a lot of words are shortened a lot of them are the exact same, so it’s easier to confuse the context. For instance “sel” could be “sinulla”, “sillä”, or beginning of “sellainen”. But if you write “sellaane” the LLM really isn’t going to be too confused. That’s why you need longer sentences to try and confuse it.
However, if you go back 10 years and more, speaking in dialects made it so translators didn’t have a snowball’s chance in hell. Even the actual Finnish book language translations were all complete garbage.
Now you can pick up context quite well even in completely obscure languages. If you don’t speak the language, you don’t know how inaccurate it is, but unless it’s something very suitable but misleading, you’re probably gonna understand that “ok this didn’t work”.
But yeah it was about 10 years ago when I had a top of the line Huawei with actual text translation from images and speech. I used to entertain my friends by showing how bad it is by just “talking” in asian languages that I sort of knew the general intonation of and maybe some sounds. Not really any words though. Chinese was clearly the worst. I could just gibber sounds and it would try to force it to be a sentence. And more often than not, the answer just sounded like some random Chinese poetry.
It didn’t even have Finnish. Finnish from actual live language has only been in a few years. (And it too sucks and governments should definitely not be using it to transcribe anything officially it’s gonna be so filled with hallucinations but already also very used).
No wörries, I do excuse you, for you know not of what you speak of. :>