Machine translations are often brought up as a gotcha whenever I criticize LLMs.
-
I have the impression that primarily anglophone people don't read as much translated literature, because so much good literature already exists in their language, so this issue may not be as familiar within that demographic. As someone who did not grow up anglophone, I can tell you there is a world of difference between a good and a bad translation even when done by humans. Machine translations are not even on the scale.
Translations are like husbands. There are beautiful translations and there are faithful translations but the beautiful ones are not faithful and the faithful ones are not beautiful.
@Gargron -
-
@Gargron would you know if you've seen a good outcome of an LLM? You'd somehow be able to identify when the LLM got it right?
I assure you you've experienced good LLM output and don't even know it. Because that's what good LLM output looks like. Indistinguishable from human output.
Your examples are perhaps false equivalencies. Take asbestos. We didn't abolish insulation. We developed better, safer insulation. We didn't stop dying food colors, we just developed safer dyes etc.
Oh LLM output always looks good, as long you don't understand what it's talking, then it looks great.
Very beautiful, very plausible.
But if you actually understand whatever it is that the LLM is talking about, then it rapidly becomes obvious that it's just spewing all the right words in a random order.
@Tekchip @Gargron -
Machine translations are often brought up as a gotcha whenever I criticize LLMs. It's worth pointing out two things: Machine translations existed decades before LLMs, and yes, machine translations are useful. However: I would never in my life read a machine translated book. Understanding what a social media post is talking about in rough terms? Sure. Literature? Absolutely not. Hell, have you ever seen machine translated subtitles? It's absolute garbage.
@Gargron funnily enough the transformer block used in most LLMs was invented for translation and TBH LLMs are much better at translation than anything before them. Not refuting your point but as machine translation goes LLMs are the best we have.
-
Machine translations are often brought up as a gotcha whenever I criticize LLMs. It's worth pointing out two things: Machine translations existed decades before LLMs, and yes, machine translations are useful. However: I would never in my life read a machine translated book. Understanding what a social media post is talking about in rough terms? Sure. Literature? Absolutely not. Hell, have you ever seen machine translated subtitles? It's absolute garbage.
Also, LLMs are making machine translations worse by adding hallucinated content into the translations:
https://www.404media.co/ai-translations-are-adding-hallucinations-to-wikipedia-articles/
-
From what I've observed, people who claim that LLMs can replace artists don't understand art, people who claim that they can replace musicians don't understand music, people who claim that they can replace writers don't understand literature, and people who claim they can replace translators don't rely on translations. If I had a button that would erase LLMs from the world but it would take machine translations away (which is a false dichotomy anyway), I would absolutely still press it.
@Gargron exactly. To me, all forms of art are still about the human connection-- "Art is a tryst, for in it maker and beholder meet." The artist is communicating something that the beholder receives. With computer-generated art, the sentience on the other end is simply not there, and any "connection" is just an illusion.
-
i do appreciate automatic subtitles extremely for hitting all my humor-chords. may they never evolve.
that being said:
i am lucky and able to read in several languages and read a lot of our bookclub books in original language. i can't count how many times i liked books that many of the others couldn't even finish their translated ones (assumedly) because the language was so poor. (and then we have those who listen to books and it totally depends on the person who was recorded. -
@Gargron we also had Concorde but it wasn’t economically viable. I mention that because I find that economic arguments seem to be heard more readily than moral arguments. (I often find that moral arguments induce temporary deafness in pro-AI people.)
@benedictc @Gargron Concorde wasn't morally viable either, both in terms of passenger safety and damage to the environment.
-
@aeva@mastodon.gamedev.place @Gargron@mastodon.social @gabboman@gabboman.xyz
ドソキーユング
lmao, love it -
@aeva@mastodon.gamedev.place @Gargron@mastodon.social @gabboman@gabboman.xyz
ドソキーユング
lmao, love it@Gargron@mastodon.social @aeva@mastodon.gamedev.place @gabboman@gabboman.xyz pretty much Dohkey Konq
-
@Gargron if asbestos was invented last year it would be inevitable, I'm afraid.
When almost all legislative power has been captured by corporatism there's not much hope we could outlaw such poisons.
-
@Gargron would you know if you've seen a good outcome of an LLM? You'd somehow be able to identify when the LLM got it right?
I assure you you've experienced good LLM output and don't even know it. Because that's what good LLM output looks like. Indistinguishable from human output.
Your examples are perhaps false equivalencies. Take asbestos. We didn't abolish insulation. We developed better, safer insulation. We didn't stop dying food colors, we just developed safer dyes etc.
-
@Gargron As an LLM would say to a translator: "All your job are belong to us".
-
From what I've observed, people who claim that LLMs can replace artists don't understand art, people who claim that they can replace musicians don't understand music, people who claim that they can replace writers don't understand literature, and people who claim they can replace translators don't rely on translations. If I had a button that would erase LLMs from the world but it would take machine translations away (which is a false dichotomy anyway), I would absolutely still press it.
@Gargron
And on the other hand, Maths people have always been saying stay the hell away from it!
https://dotnet.social/@SmartmanApps/116000100388648367 -
Machine translations are often brought up as a gotcha whenever I criticize LLMs. It's worth pointing out two things: Machine translations existed decades before LLMs, and yes, machine translations are useful. However: I would never in my life read a machine translated book. Understanding what a social media post is talking about in rough terms? Sure. Literature? Absolutely not. Hell, have you ever seen machine translated subtitles? It's absolute garbage.
@Gargron I've read machine translated versions of Beowulf, which I've read in various translations dozens of times. It's my favorite story of all time. And Old English, like Old Norse is a very descriptive language with hidden meaning behind it. And these machines loose those linquistic nuances. You can tell when a medieval text is machine translated...because they're not all that exciting. And Beowulf is exciting!
-
Technology is not inevitable. We've decided not to have asbestos in our walls, lead in our pipes, or carginogenic chemicals in our food. (If you're going to argue that it's not everywhere, where would you rather live?) We could just not do LLMs. It's allowed.
@Gargron The problem is that the pieces of shit peddling LLMs have convinced the political class that it's a race to technological supremacy and that any nation that bans them will be left behind. When in reality they're more like opium.
-
@virgilpierce @Gargron
There's an old joke from the 1960s about machine translation of the saying "the spirit is willing, but the flesh is weak" from English to Russian and then back again.
The result was "the vodka is good but the meat is rotten."@gdinwiddie I quoted this a number of times over the past few decades
(I remembered it as "the spirit is strong", BTW) @virgilpierce @Gargron -
Machine translations are often brought up as a gotcha whenever I criticize LLMs. It's worth pointing out two things: Machine translations existed decades before LLMs, and yes, machine translations are useful. However: I would never in my life read a machine translated book. Understanding what a social media post is talking about in rough terms? Sure. Literature? Absolutely not. Hell, have you ever seen machine translated subtitles? It's absolute garbage.
@Gargron Some titles are perfectly constructed pearls of wisdom and insight. LLMs wouldn't have a clue.
-
@Gargron But it seems that LLMs are here to stay. This time, it doesn't seem to be just a passing fad. There is a lot of investment involved.
-
Also, LLMs are making machine translations worse by adding hallucinated content into the translations:
https://www.404media.co/ai-translations-are-adding-hallucinations-to-wikipedia-articles/
@Wally I mostly use machine translations when trying to read 16th century Danish. This is kind of a niche language variant to be reading so I don't expect them to be very good at it and that's fine, I'm just trying to get the gist usually.
Up to early 2025, Google Translate(*) was really useful for this. It would translate it more or less as if it was modern Danish, and every time it came across a word it didn't know(**), it would leave it untranslated. I could then either figure it out myself (from context, or prior experience, or "oh yeah uu = w" or whatever), or look it up in a contemporary dictionary, or decide it didn't matter for my purposes.
Then around April 2025 I started noticing it was translating *all* the words, and the resulting sentences sounded really clean and smooth. It's just they were sometimes wrong. Because now when it didn't know a word, it was putting in whatever English word it thought would be most statistically probable in the context. (At the same time, Highly Coincidentally, longer translations now made it prone to breaking down completely into LLM-gibberish.)
So since then I haven't been able to trust *any* part of its translations that I can't verify with my own knowledge of the language.
(*) I've tried various other products including open source ones, just haven't found anything as competent for my purposes.
(**) I'm using 'know' and the like as shorthand only, I know it doesn't know anything.
