I remember watching a film online that had AI generated subtitles. it was shit because the subtitles said vaguely the same thing but were still very different to what was actually said. and both the film and the subtitles were in English. can’t imagine how bad AI translations will be. also what’s wrong with getting a translator, who can understand context?
In my experience LLM get vastly better results that traditional translation software.
Also google translate is not traditionally suited for long coherent text. One particular issue is tone, proper translation takes into account not only the worlds but the tone and the subject is being treated. Google translate cannot take that into account, with an llm the user can tweak the tone to match the tone of the original text with better accuracy.
And, anyway google translate if it’s not already using llm for translation will soon. Results are just better. It’s one of the tasks that language models are actually good for.
Anyhow, what’s the issue if an automatic translation is done using one software or other? Just use whatever gives best results and it’s more convenient for the developer.
I don’t have an issue, it was just a question.
I don’t ever translate anything but a few words myself.
My assumption was that a dedicated tool would have done a better job, but you have good points about tone and coherence of long text (and possibly even across many promps of the translation).
Just so we’re clear, the first pass of localization of every game you’ve played in the past decade has been machine-generated.
Which is not to say the final product was, people would then go over the whole text database and change it as needed, but it’s been frequent practice for a while for things like subtitles and translations to start from a machine generated first draft, not just in videogames but in media in general. People are turning around 24h localization for TV in some places, it’s pretty nuts.
Machine generated voices are also very standard as placeholders. I’m… kinda surprised nobody has slipped up on that post-AI panic, although I guess historically nobody noticed when you didn’t clean up a machine-translated subtitle, but people got good at ensuring all your VO lines got VOd because you definitely notice those.
As with a lot of the rest of the AI panic, I’m confused about the boundaries here. I mean, Google Translate has used machine learning for a long time, as have most machine translation engines. The robot voices that were used as placeholders up until a few years ago would probably be fine if one slipped up, but newer games often use very natural-sounding placeholders, so if one of those slips I imagine it’d be a bit of drama.
I guess I don’t know what “AI generated” means anymore.
I haven’t bumped into the offending text in the game (yet), but I’m playing it in English, so I guess I wouldn’t have anyway? Neither the article nor the disclosure are very clear.
That said, the game is pretty good, if anybody cares.
historically nobody noticed when you didn’t clean up a machine-translated subtitle
I don’t know about that, it’s super noticeable when that happens, it’s just that it mostly affects languages other than English, so it did not get noticed by Western media unless there is a review bombing campaign after a particularly atrocious localization
As a non-native English speaker, let me tell you, terrible localization was very much a thing that happened well before machine translation, so that by itself (and more subtle typos or one-off errors) was definitely not enough to infer that someone had forgotten to fix a machine-translated line once.
You can definitely tell when something has been machine-translated and not fixed, but the real challenge is lack of context. This leads to nonsensical localization even today, whether it’s human or automated, especially in crowdsourced localizations, which are frequent in open source software. I contribute to some on occassion and maaaan, do I wish well intentioned people in that space would stop contributing to projects they don’t use/lines they haven’t seen in situ.
Ok? It was a temporary voice file that the devs forgot to remove or replace. And people immediately screamed that Blizzard is trying to sneak AI into the game.
I hadn’t clicked through to the Reddit thing (for obvious reasons). The example in the article proper is in a Portuguese subtitle, but now that you pointed me at it and I did check the Reddit thread… well, that text is not legible in game unless you really try, so yeah, I hadn’t read it. I’m guessing that’s the only English instance?
Shameful, but this is the state of modern game developers. Scrap every possible avenue of paying your workers a living wage while surrendering to all latest failure tech fads.
The examples people found include a monitor showing random technical text that someone asked a LLM to write (presumably the writer who goofed is getting paid) and some localized subtitles that were left with a machine localization (the rest of the localization was contracted out).
Even assuming a bunch of other stuff in the game was AI generated and just went undetected, which is likely, if it’s all iterations on what people noticed it definitely doesn’t fit your description.
As the Industrial Revolution began, workers naturally worried about being displaced by increasingly efficient machines. But the Luddites themselves “were totally fine with machines,” says Kevin Binfield, editor of the 2004 collection Writings of the Luddites. They confined their attacks to manufacturers who used machines in what they called “a fraudulent and deceitful manner” to get around standard labor practices. “They just wanted machines that made high-quality goods,” says Binfield, “and they wanted these machines to be run by workers who had gone through an apprenticeship and got paid decent wages. Those were their only concerns.”
Me and the boys being worried for our jobs as automation stretches onwards and just wanting some level of guarantee that good paying jobs will still be available 😭
For the record, the word as a general noun is widely recognized to mean what everybody thinks it means:
Luddite
noun
Ludd·ite ˈlə-ˌdīt
: one of a group of early 19th century English workmen destroying laborsaving machinery as a protest
broadly : one who is opposed to especially technological change
One of the weirder annoyances of the AI moral panic is how often you see this spiral of pedantry about the historical luddites whenever someone brings up the word as a pejorative.
I mean, fair rhetorical play, I suppose, in that it creates a very good incentive to not bring it up at all. If the goal was to avoid being called a luddite as an insult or as shorthand for dismissing AI criticism as outright technophobia I suppose that is mission accomplished, disingenuous as it is.
It is also correct that someone disagreeing with me can be doing so because of a moral panic. Our agreement is entirely disconnected to whether there is a moral panic at play or not.
For the record, I think “AI” is profoundly problematic in multiple ways.
This is also unrelated to whether there is a moral panic about it. Which there absolutely is.
Showing that even a small non-issue use of AI will be detected is a pretty strong incentive for other games to disclose that willingly. Otherwise, why would they admit to it if no one can tell? Morals??? 😂😂😂
Gamers who get mad having to wait for company logos to show while games boot: They have to disclose every piece of software they use to make every game! I want my games 100% hand crafted and bespoke. I want to sense the life people spent meticulously crafting mudsplat_texture_1 - mudsplat_texture_500. Also no crunch (and no bugs, obviously)
It’s just reality. Selecting a bunch of textures nobody has seen before through AI but hand crafting the rest of the game would force them to wear the scarlet letters on their front, and open them up to brigading by the brainwashed NeoLuddite mob. That move by steam to appease the pitchfork masses is pretty heavy handed.
even if I accept your premise of a “brainwashed NeoLuddite mob,” you’re still wrong on the simple principle of arguing against a company properly labeling what their product is or how it was made.
When a company makes a product completely by hand, but uses AI for a few translations it gets the same label as a game pumped out of Grok and uploaded untouched. That label is misleading and punitive, not informative.
You are not logged in. However you can subscribe from another Fediverse account, for example Lemmy or Mastodon. To do this, paste the following into the search field of your instance: [email protected]
Welcome to the largest gaming community on Lemmy! Discussion for all kinds of games. Video games, tabletop games, card games etc.
Rules
1. Submissions have to be related to games
Video games, tabletop, or otherwise. Posts not related to games will be deleted.
This community is focused on games, of all kinds. Any news item or discussion should be related to gaming in some way.
2. No bigotry or harassment, be civil
No bigotry, hardline stance. Try not to get too heated when entering into a discussion or debate.
We are here to talk and discuss about one of our passions, not fight or be exposed to hate. Posts or responses that are hateful will be deleted to keep the atmosphere good. If repeatedly violated, not only will the comment be deleted but a ban will be handed out as well. We judge each case individually.
3. No excessive self-promotion
Try to keep it to 10% self-promotion / 90% other stuff in your post history.
This is to prevent people from posting for the sole purpose of promoting their own website or social media account.
4. Stay on-topic; no memes, funny videos, giveaways, reposts, or low-effort posts
This community is mostly for discussion and news. Remember to search for the thing you’re submitting before posting to see if it’s already been posted.
We want to keep the quality of posts high. Therefore, memes, funny videos, low-effort posts and reposts are not allowed. We prohibit giveaways because we cannot be sure that the person holding the giveaway will actually do what they promise.
5. Mark Spoilers and NSFW
Make sure to mark your stuff or it may be removed.
No one wants to be spoiled. Therefore, always mark spoilers. Similarly mark NSFW, in case anyone is browsing in a public space or at work.
6. No linking to piracy
Don’t share it here, there are other places to find it. Discussion of piracy is fine.
We don’t want us moderators or the admins of lemmy.world to get in trouble for linking to piracy. Therefore, any link to piracy will be removed. Discussion of it is of course allowed.
I remember watching a film online that had AI generated subtitles. it was shit because the subtitles said vaguely the same thing but were still very different to what was actually said. and both the film and the subtitles were in English. can’t imagine how bad AI translations will be. also what’s wrong with getting a translator, who can understand context?
Why would they use LLMs for translation over dedicated translator (apps like Google translate)? Is this common now?
In my experience LLM get vastly better results that traditional translation software.
Also google translate is not traditionally suited for long coherent text. One particular issue is tone, proper translation takes into account not only the worlds but the tone and the subject is being treated. Google translate cannot take that into account, with an llm the user can tweak the tone to match the tone of the original text with better accuracy.
And, anyway google translate if it’s not already using llm for translation will soon. Results are just better. It’s one of the tasks that language models are actually good for.
Anyhow, what’s the issue if an automatic translation is done using one software or other? Just use whatever gives best results and it’s more convenient for the developer.
I don’t have an issue, it was just a question.
I don’t ever translate anything but a few words myself.
My assumption was that a dedicated tool would have done a better job, but you have good points about tone and coherence of long text (and possibly even across many promps of the translation).
I work with multiple languages and I haven’t heard of LLMs being used instead of Deeply/Google translate.
Don’t really see the reason to switch to LLMs.
Just so we’re clear, the first pass of localization of every game you’ve played in the past decade has been machine-generated.
Which is not to say the final product was, people would then go over the whole text database and change it as needed, but it’s been frequent practice for a while for things like subtitles and translations to start from a machine generated first draft, not just in videogames but in media in general. People are turning around 24h localization for TV in some places, it’s pretty nuts.
Machine generated voices are also very standard as placeholders. I’m… kinda surprised nobody has slipped up on that post-AI panic, although I guess historically nobody noticed when you didn’t clean up a machine-translated subtitle, but people got good at ensuring all your VO lines got VOd because you definitely notice those.
As with a lot of the rest of the AI panic, I’m confused about the boundaries here. I mean, Google Translate has used machine learning for a long time, as have most machine translation engines. The robot voices that were used as placeholders up until a few years ago would probably be fine if one slipped up, but newer games often use very natural-sounding placeholders, so if one of those slips I imagine it’d be a bit of drama.
I guess I don’t know what “AI generated” means anymore.
I haven’t bumped into the offending text in the game (yet), but I’m playing it in English, so I guess I wouldn’t have anyway? Neither the article nor the disclosure are very clear.
That said, the game is pretty good, if anybody cares.
I don’t know about that, it’s super noticeable when that happens, it’s just that it mostly affects languages other than English, so it did not get noticed by Western media unless there is a review bombing campaign after a particularly atrocious localization
As a non-native English speaker, let me tell you, terrible localization was very much a thing that happened well before machine translation, so that by itself (and more subtle typos or one-off errors) was definitely not enough to infer that someone had forgotten to fix a machine-translated line once.
You can definitely tell when something has been machine-translated and not fixed, but the real challenge is lack of context. This leads to nonsensical localization even today, whether it’s human or automated, especially in crowdsourced localizations, which are frequent in open source software. I contribute to some on occassion and maaaan, do I wish well intentioned people in that space would stop contributing to projects they don’t use/lines they haven’t seen in situ.
Diablo 4 had this recently, where an obviously Microsoft Sam like robot voice made it through, and a few people lost their minds.
I thought MS Sam was an accessibility feature, that you can enable and disable text-to-speech. Did they use that for NPCs voice-overs?
Ok? It was a temporary voice file that the devs forgot to remove or replace. And people immediately screamed that Blizzard is trying to sneak AI into the game.
Oh, I did not know that. Thanks for explaining it.
The text in the screenshot in the reddit post they link is in English
I hadn’t clicked through to the Reddit thing (for obvious reasons). The example in the article proper is in a Portuguese subtitle, but now that you pointed me at it and I did check the Reddit thread… well, that text is not legible in game unless you really try, so yeah, I hadn’t read it. I’m guessing that’s the only English instance?
Shameful, but this is the state of modern game developers. Scrap every possible avenue of paying your workers a living wage while surrendering to all latest failure tech fads.
Neither of those things happened here.
The examples people found include a monitor showing random technical text that someone asked a LLM to write (presumably the writer who goofed is getting paid) and some localized subtitles that were left with a machine localization (the rest of the localization was contracted out).
Even assuming a bunch of other stuff in the game was AI generated and just went undetected, which is likely, if it’s all iterations on what people noticed it definitely doesn’t fit your description.
What an insanely non-issue to clutch one’s pearls at… Between Luddites and Trumpites, this is the worst timeline…
[Really appropriate example, actually. ](https://www.smithsonianmag.com/history/what-the-luddites-really-fought-against-264412/]
Me and the boys being worried for our jobs as automation stretches onwards and just wanting some level of guarantee that good paying jobs will still be available 😭
For the record, the word as a general noun is widely recognized to mean what everybody thinks it means:
One of the weirder annoyances of the AI moral panic is how often you see this spiral of pedantry about the historical luddites whenever someone brings up the word as a pejorative.
I mean, fair rhetorical play, I suppose, in that it creates a very good incentive to not bring it up at all. If the goal was to avoid being called a luddite as an insult or as shorthand for dismissing AI criticism as outright technophobia I suppose that is mission accomplished, disingenuous as it is.
Someone disagreeing with you =/= moral panic
But does them disagreeing with you preclude a moral panic?
of course not
That is correct.
It is also correct that someone disagreeing with me can be doing so because of a moral panic. Our agreement is entirely disconnected to whether there is a moral panic at play or not.
For the record, I think “AI” is profoundly problematic in multiple ways.
This is also unrelated to whether there is a moral panic about it. Which there absolutely is.
long winded way to say your objections are logical and sound while everyone else is just having a panic, you little moralizer you.
Well, no, it’s a concise way to say some objections are logical and sound and some are stemming from a moral panic.
Whether I agree with the objections on each camp is, again, irrelevant.
I disagree with some of the non-moral panic objections, too, and I’m happy to have that conversation.
Four possible types of objections in this scenario, if you want to be “logical” about it:
I think there aren’t any in that last group, but there are certainly at least some objections in all other three.
Ah, the always welcome “and so what?” comment, with a side of name calling for some extra spice.
Don’t forget the obligatory smug response from somebody who feels they have moral superiority.
Showing that even a small non-issue use of AI will be detected is a pretty strong incentive for other games to disclose that willingly. Otherwise, why would they admit to it if no one can tell? Morals??? 😂😂😂
Why should they “admit” to it when NeoLuddites with pitchforks and no ability to reason lurk around every corner?
A substantial part of the market not wanting AI in their products is actually a great reason for them to disclose when it is or isn’t used
Like I’m not the biggest fan of gen ai but a generic computer screen feels like a good use case for filler text.
It wouldn’t be lemmy if we weren’t always in a moral size measuring contest.
Gamers who get mad having to wait for company logos to show while games boot: They have to disclose every piece of software they use to make every game! I want my games 100% hand crafted and bespoke. I want to sense the life people spent meticulously crafting mudsplat_texture_1 - mudsplat_texture_500. Also no crunch (and no bugs, obviously)
Nice strawman Dorothy
It’s just reality. Selecting a bunch of textures nobody has seen before through AI but hand crafting the rest of the game would force them to wear the scarlet letters on their front, and open them up to brigading by the brainwashed NeoLuddite mob. That move by steam to appease the pitchfork masses is pretty heavy handed.
even if I accept your premise of a “brainwashed NeoLuddite mob,” you’re still wrong on the simple principle of arguing against a company properly labeling what their product is or how it was made.
When a company makes a product completely by hand, but uses AI for a few translations it gets the same label as a game pumped out of Grok and uploaded untouched. That label is misleading and punitive, not informative.
speaking about hypothetical events as if they were real is unhinged
you’re unhinged