{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreidfqt5ltzg5xafv3pkr5wie54shb2la4ezemy3cnwf7vnuymptd3m",
    "uri": "at://did:plc:npppinc2x6on5fmrcemn2p5o/app.bsky.feed.post/3mfoo3dtx6tv2"
  },
  "path": "/post/809518508376506368",
  "publishedAt": "2026-02-25T10:56:50.000Z",
  "site": "https://tumblr.sztupy.hu",
  "tags": [
    "voyaging-too",
    "eikaronswelt"
  ],
  "textContent": "voyaging-too:\n\n> eikaronswelt:\n>\n>> voyaging-too:\n>>\n>>> voyaging-too:\n>>>\n>>>> I don’t want to be overly dramatic and overly negative about the AI translations I’ve been working with. They are bad, yes, but I don’t want to overstate their badness because that would obscure the specific points I’m making. Some AI translations, the best AI translations, are not that bad. Some of them are still bad to the point of being unusable, but others are better. They’re not good, but they’re mostly serviceable, and it’s extremely impressive that a machine can come up with something serviceable, something comparable to the work of a very mediocre human translator.\n>>>>\n>>>> A client who hires a subpar translator who accepts being underpaid, in order to avoid paying professional rates for a professional, is getting subpar work. A client who uses AI to get work cheap and fast is getting worse than subpar work. But AI is getting better, it might soon be at the point where laymen can’t tell the difference, and then, using AI instead of paying a human will mostly be a labour rights issue, and that’s a far thornier question. (Note that I’m not talking about using AI translators to read something for yourself, or to communicate in your daily life: I’m talking about AI translation for publication, using AI for something you expect other people to pay money for.)\n>>>>\n>>>> My actual point about AI translation is that even when it’s fairly good, when it makes few errors and conveys the message intelligibly, it lacks something. I’m not talking about heart and soul here, nothing to do with some intangible human quality: I’m talking about specificity. AI works with great averages, and so it automatically irons out nuance. If you write something unusual, AI will assume it’s an error, instead of an intentionally unusual statement. This is regression to the mean, and based on the texts I’m working with, it’s an Anglophone, American mean. If you say something that’s true of 1980s Hungary, it might slightly alter the sentence to “make sense” for 1980s US. Some alterations are factual, these are more serious errors but also easier to edit out. But other things are harder to catch, slight shifts in tone and valence, an erasure of the original, specific, non-American perspective, and the end result is a text that _doesn ’t have anything wrong with it,_ but is markedly simpler and dumber than it should be. And flattening complex, knotty, peripheral perspectives into something closer to a monoculture is, in the long term, intellectually devastating.\n>>>\n>>> I want to emphasise that what I described here, and what I was trying to be measured and not-overly-negative about is the best case scenario, and the text I was editing then was fairly close to that impressive yet depressing best case scenario.\n>>>\n>>> Yesterday and today I was editing an AI translation so bad, so so bad, I don’t even know where to start. It translated fried eggs as mirrored eggs. It misjudged homophones and translated nails as angles, slideshows as students. It fell on its ass the first time it saw a common, widely used slang term. It translated proper nouns it shouldn’t have, and it didn’t translate proper nouns that have an official translation and their own wiki page. It decided that a man with the surname Erdély was literally the geographical area of Transylvania. It failed to notice the fairly important difference between “was a labourer” and “was deported to a labour camp.” Surprisingly, there was only one incident of misgendering. I can’t be too negative about this, this is pure shit. If your sole aim was to get a general idea of what the text was about, it may be good enough, although still not great. If you want this to be a translation, it’s not.\n>>\n>> Ok, I’m assuming the text was in German because Spiegeleier… Did the author make a joke of some sort by writing sth like “gespiegelte Eier” instead of Spiegeleier otherwise it’s wild this didn’t get translated correctly. I mean that is not in any way, shape or form an unusual word. But then neither is “Nägel” and I’m at a loss for what could have possibly happened with Slideshow and students. Presumably you can’t just post it here but ngl I’m dying to see the full text with the bad translation\n>\n> To clarify: the original text was in Hungarian, we also use the same term, Spiegeleier/tükörtojás/mirror-egg to describe fried eggs, especially sunny side up. Fairly sure it’s one of our many borrowings from our Habsburg neighbour-overlords.\n>\n> Nails and angles are both “szög”, but context should have clarified the difference. A slide is “dia”, slides plural is “diák”, student singular is also “diák”, but the word was used in a sentence, in a paragraph, in a text talking about slideshows, and not students. I’m not saying a human being would never have made these mistakes, I’m saying a human would have had to be spectacularly drunk to make them.\n>\n> You’re right: none of these are in any way, shape or form unusual words. It’s wild that they didn’t get translated correctly. One of the problems with machine translation is the loss of subtle nuances and subjective linguistic and intellectual complexity, another problem with machine translation is just plain garbage output, like translating tükörtojás correctly as fried egg, and then a few paragraphs later parsing the same term as mirrored egg, because why the fuck not.",
  "title": "I don’t want to be overly dramatic and overly negative about the AI translations I’ve been working…"
}