submitted 2 months ago by [email protected] to c/[email protected]
top 50 comments
sorted by: hot top controversial new old
[-] [email protected] 82 points 2 months ago* (last edited 2 months ago)

I am a translator. Some decades ago the language industry introduced MT - some kind of precursor of LLM. The prices of translation jobs didn't change, and translators didn't lose their work entirely. But gradually we were offered more and more MTPE (euphemism for fixing the robot's shit) jobs, for a lower rate. Many older colleagues stayed with the few remaining translation jobs, young people starting out became "MTPE editors". These days there are a few translation jobs, many MTPE jobs, and more and more jobs in "AI output rating" - and the new generation will be working as an "AI linguistic assistant" or other such barbarity for even less money.

The tech isn't necessarily bad in itself, but what we have to wake up to is that tech is used to pay each generation after us a little less. We have to resist this and demand fair pay for fair work always - no matter if they want to call it 'translation', 'AI output review' or 'ertdfg sfdgs' - it has a price, and this price has to respect our dignity and enable a healthy life for us language workers and all other workers.

[-] [email protected] 19 points 2 months ago* (last edited 2 months ago)

The best part is, the MTPE workers' output is 100% going to get fed back into the algorithms, so it's only a matter of time before the average error rate of the models is good enough that there's no real reason to pay anyone to look at it.

People are, I think, overly optimistic that AI won't eventually take their job. Why the hell do people think their boss wants to pay them plus pay for an AI? When they can, they'll just switch as much as they can over to AI. We have a quantifiable error rate/range of error rates for most tasks, so all they have to do is create a model with an average error rate lower than, say, most people, and the case for employing us to even review their output goes out the window. It's not like humans don't make mistakes, so in reality, we 100% are in competition with AI for every task category. AI doesn't have to be perfect to make us obsolete, it just has to be overall cheaper with an "acceptable" output quality.

In fact, businesses often accept tradeoffs in quality in exchange for cost savings, so AI doesn't even have to get better at something to replace us. If it costs 1% and is 80% as good, and that 20% drop in quality isn't enough to affect their bottom line, you can bet your ass that humans won't be doing that job anymore. I've already seen comments from copy writers about how they lost clients to ChatGPT on exactly these grounds. ChatGPT.

Technology development takes a long time, I'm thinking 30-50 years out here, but the point is, this is a different kind of technological development than earlier ones. Future generations might, in fact, not have jobs at all.

[-] [email protected] 5 points 2 months ago

Yeah at that point they will likely push us into a situation where violence with capital is inevitable. I’m not super in agreement with Marx but he had some points here

[-] [email protected] 1 points 2 months ago

The best part is, the MTPE workers’ output is 100% going to get fed back into the algorithms, so it’s only a matter of time before the average error rate of the models is good enough that there’s no real reason to pay anyone to look at it.

Not quite, yes and no. It will be ever so slightly off, and the slightly off gets fed back, and language will change. It is already happening since CAT tools introduced segmentation - which means texts were segmented, and thus translated, at sentence level. Where a human translator in front of an unaltered text would have joined some sentences and separated others in the translated text, one tends to stick to the segmentation when working in a CAT tool, and MT- and CAT-translated German sounds almost, but not quite, like human German - remember that beverage machine making tea in the Hitchhiker's Guide? Only we got used to drinking the stuff and reading the texts, what can we do.

And now you have the added feature of MT flavoured translated texts. I for my part really enjoy handmade things these days, fuck AI. I thought AI generated images were funny at first, then understood the computation needs involved, the intellectual property stolen and put to work for corporations, the implications for many digital workers ... it's just more awful than fun honestly and should just fuck off back into the arseholes of the thieving scammers who invented it.

[-] [email protected] 13 points 2 months ago* (last edited 2 months ago)

Yeah that's pretty consistent with my expectations. A lot of work will transition into fixing the robots mistakes. So we'd be ceding the interesting, more creatively challenging aspects of our jobs to AIs and turning into data janitors. And that would only last as long as we'd be necessary. They'd hammer out the details making that janitor work eventually disappear.

I do design and illustration and it'd kind of be like telling me "Well we don't need you to illustrate this stuff anymore, but Midjourney still draws shitty hands with too many fingers. So your job now is to fix those hands." That is not what I came here to do and that does not provide the fulfillment I seek from a line of work. And following that analogy, Midjourney will eventually make flawless hands and I'd be out of a job.

Fortunately right now AI cannot hit a specific design/illustration brief to the consistent standards my projects require, nor iterate on a project based on specific and vague client feedback. So I still have work for now, but I see the writing on the wall. I'm always surprised other people don't see that writing too.

This whole thing is going to make an insane chasm of the wealth equality divide we already have.

[-] [email protected] 4 points 2 months ago

Have you not noticed any effect from midjourney on your line of work yet?

[-] [email protected] 7 points 2 months ago* (last edited 2 months ago)

Yeah for sure. Mostly indirectly. I know a few people in my line of work who lost jobs because the client decided to just use AI to generate something.

I've also seen a number of examples of publications using AI images for editorial pieces which absolutely used to be paying jobs. For example this Atlantic article on Alex Jones. An actual person would have been paid to do a piece like this before AI came around.

And also there was the San Francisco ballet that did a bunch of their Nutcracker promo campaign art with AI stuff last year. They had traditionally used artists and photographers for years to do key pieces for their promo materials.

And as far as I am personally concerned, I've seen a marked slump in the volume of work inquiries I've been getting in the last year. I've been fortunate enough to remain fully booked and in the past just had to turn down a lot of work, but right now I'm getting about half as many inbound inquiries as I would have even a year ago. Hard to pin that on any one thing but I am sure AI is a factor. I'd be lying if I said that there haven't been a number of my jobs over the years that couldn't have been done with one of these AI models and a little trial and error.

I've also had a few clients now send me Midjourney stuff and basically want me to replicate it but make it work for whatever thing it was they were needing artwork for. So right there, that's all the fun problem solving and artistic exploration out the window and it's basically a case of "fix the robot's thing." It's pretty depressing.

I'd be mostly fine with the robots doing away with all of our jobs if it meant we were heading into some post-work utopia where we got to just spend time doing the things that really matter to us, but that's almost definitely not where this is going. All the windfalls will go to the top, the jobs will be less interesting, and wages will be depressed.

[-] [email protected] 2 points 2 months ago

Your last paragraph really sums up my feelings on all of this. Work becoming easier and less needing to be done should be a good thing but we’re handling it terribly as a society

[-] [email protected] 3 points 2 months ago

I am literally fixing punctuation all day that AI is too stupid to pick up. But it translates whole paragraphs most beautifully. I spend most of my working day in some state of dissociation, with an occasional laugh as I watch what I thought was civilization crumble before my eyes - as always, my work software wanted an update before starting a new project, and as ever more often, didn't work after the update. Nobody gives a shit about quality anymore and I guess we're all on drugs or suffer the consequences of long Covid, or both.

[-] [email protected] 3 points 2 months ago

Yeah I feel like it must have really done a number on the field of translation. Also voice over work at the low to mid budget is probably done for with what those voice AI models can do now. It's a sad state of affairs and it's disheartening to see so many people cheer it on without caveats.

[-] [email protected] 2 points 2 months ago

Tell me about it! I'm an audiobook narrator and I jumped on the AI bandwagon because my job as is, is bound to disappear. Now I can offer audiobooks narrated with my cloned voice for my clients with low budgets.

Audible doesn't accept ai narrated books as of yet, but it's just a matter of time.

Interestingly though, none of my clients went the AI road yet and still prefer to pay me rather than 3 times less for my AI voice. I bet that won't last though.

I'm also looking into completely changing field. How about healthcare, I'm sure they'll never stop needing and/or abusing those in this field.....

[-] [email protected] 1 points 2 months ago

It's of course heavily hyped. No matter that the AI they are hyping for translation now isn't much different from the MT we are using since more than a decade.

Btw I like the MT with reservations as it saves my hands from typing a lot. And I have been picky in choosing clients and negotiating my services, so I can't complain personally. But I would like to help organize online language workers in some way, and make sure AI doesn't fuck up quality even further, and demand reasonable rates. Especially for those poor folks in cheap translation mills (forgot their names). I also have heard of rates like 0.03€ doing translations for a (South) EU government, work for which usually a language degree is necessary, yet people accept these laughable shit rates.

[-] [email protected] 8 points 2 months ago

Sounds to me like you need a union...

[-] [email protected] 6 points 2 months ago

We used to and still have translators' associations, but most of them are stuck in the past. I was proving my skills as a translator by sitting at a desk and handwriting my translation while looking up stuff in physical dictionaries. They probably imagine that we are sitting in an office waiting for clients to walk in and hand us sheets of paper. It's still like this for a few of us, but the vast majority works as typing monkeys part of huge international teams and churns out translations by the meter, and can't afford neither the overpriced exam fees nor the inflated membership fees of organizations who do very little to support the positions of online translators.

So yes, we need a union. But I think it should be international, and best include all digital workers. We are the burger flippers of the digital world and deserve a living wage.

Now, as to AI, I would say the problem is that it's wasteful computation-wise, and that's why I'd rather not have it. I very much value reading texts by actual people, and look at images drawn by actual people and I am willing to pay for that. I want to use hand-knitted garments, hand-woven baskets and rugs, and not have sad people sit in factories 12 hours a day just so I can afford cheap plastic gadgets instead. So the other part of this would be to refuse consuming the cheap imitation of reality they offer after stealing everyone's works. Go treat yourself to the best and most beautiful, done by someone with passion and love for their work. Don't consume trash.

The problem in the case of translators, other digital workers, and unions, however is not really about AI versus brain, machine versus hand. It's about an economic system that forces us to work all day so we can survive. If you have to flip burgers, translate, dig potatos, play the violin 8 hours a day 5 days a week to survive, that's too much. Stop. Demand better.

[-] [email protected] 5 points 2 months ago

Yes, they are being paid less. Technology is supposed to free us from mundane tasks and make things cheaper. However, neither one of these occurs because of greed and poor management. So in reality if technology is paying us less, it should also be making things cheaper for us. There is clearly a discrepancy and we should demand for lower prices or higher wages, or both as a compromise. The compromise cannot be lower wages and higher prices, that is economically destructive and fuels greed and class wars.

[-] [email protected] -3 points 2 months ago

fair pay for fair work

Sure but what's fair? As you described, the work did change considerably. Translating from scratch is much more work and also much harder than fixing a mostly ok output. It would not be fair to pay both jobs the same amount since the latter can be done by people with less expertise/education.

Eventually, AI output won't need any human editing at all. What then? Resisting change driven by technology is understandable from the individual perspective but it has always been doomed to fail. You know that "computer" used to be a job title?

[-] [email protected] 11 points 2 months ago

If 10 farmers can make enough to feed 100 people, and new tech comes out that makes it possible for 5 farmers to make enough to feed 100 people, the ideal scenario is that now all 10 farmers should only have to work half as much. What usually ends up happening is that half the farmers are laid off so the boss at the top can pocket the extra money.

This is how we end up with enough resources to feed, clothe and house everyone but still have people living in poverty. Because the system is no longer designed to provide for people, it's meant to make profit for capitalists. It makes technological progress a negative instead of the positive of should be.

[-] [email protected] 9 points 2 months ago

When our wealthy are legitimately discussing a trip to Mars, fair pay is about whatever the local McDonald's charges for a double quarter pounder x5000, per year. After taxes.

Don't ask, but check for yourself!

[-] [email protected] 2 points 2 months ago

Erm how much is than in sheep or potatos? I don't do McDonald's maths

[-] [email protected] 2 points 2 months ago

Well, that depends on the sheep.

Somewhere between like 50 and 500 sheep.

About 125 short tons of potatoes.

[-] [email protected] 6 points 2 months ago

It's logical in a capitalistic sense. Yet it's arguable if that's how it's supposed to be. With all these industrialization, automation, now LLMs, we end up working even more to survive. If not for unions, progressives counteracting it, it could be even worse. Isn't it a regression instead of a progress? Why can't they, at least, start to work less with the same pay so we all end up here somewhen? Isn't that what everyone wants in the future?

Why exactly correcting the text after an AI requires less experience? Main engineer isn't paid less than his subordinates because they don't plan every wall socket themselves, it's an opposite, their experience and competence lets them lead the project and ensure it's up to stantards. They put their personal responsibility for the work their team put together.

I'm not a native English speaker and one of the reasons I started to learn it was because my local tranlations sucked ass. In the media, in books. Sometimes I could see the remains of a mistranslated english idiom that a human translator just didn't recognize. And that's just entertainment, and a bored person who dgaf. AI is just like that. It can't care, it doesn't dig into context, it doesn't intentionally choose what to write, it can't proofread itself. Imagine trusting more important cases like world diplomacy to someone who is just aproximately right, a workbook to someone who pick terminology at random and constantly changes it, a loveletter to an automated SEO optimizer. It can help you grasp the basics of what is said, that's all.

While professional translation is the Craft. And long before the first computer, different prominent authors competed with each other with their own translations of classic and well-known texts, these all got studied and compared ad nauseum, because it's an open question how to do it better. Academics constantly argue if old names for things still fit them, they can start a feud over a slight difference in their definitions you can't smell without 30 years in a field. And instead of mentioning the Bible that had exiles and bloodsheds started over these two, I'd put there our hated TikTok that makes billions of users by making their language of images so effective it's intoxicating. Thus I insist that language fucking matters.

And although in the beginning of my rant I stated I found many mistakes in translations, these helped me understand how much it takes to decode something right. How it's easy to fail it. To appreciate how much effort and soul goes into that, even if it's just correcting.

Your dismissal of their value could be a good trolling tho, if only it was. But it seems your way of seeing that subject may be too popular in masses and obviously profitable to the moneymakers. So perceive that not as a personal reply, but just me letting a steam off for once.

[-] [email protected] 3 points 2 months ago

Technological progress is okay if it is

  1. ecologically sustainable
  2. in the hands of the public, not a few corporations.

And AI fails for both.

[-] [email protected] 56 points 2 months ago

You won't lose your job, you'll just become more of a peasant.

[-] [email protected] 21 points 2 months ago


[-] [email protected] 20 points 2 months ago

The conclusions about the EU job market don't necessarily translate to the US, the researchers caution, noting that their results contrast with previous findings from American researchers.

Research on the effect of AI on jobs in the US has indicated that generative AI could cause 2.4 million job losses by 2030. Large tech players like IBM have also cited AI as a reason to lay off thousands of people whose jobs could be replaced by software.

[-] [email protected] 16 points 2 months ago

Now you can work two jobs!

[-] [email protected] 2 points 2 months ago

For the salary of one!

[-] [email protected] 3 points 2 months ago

They will still work you for everything your worth, they're just paying less for your labor.

load more comments
view more: next ›
this post was submitted on 30 Nov 2023
176 points (95.4% liked)


51484 readers
5416 users here now

This is a most excellent place for technology news and articles.

Our Rules

  1. Follow the lemmy.world rules.
  2. Only tech related content.
  3. Be excellent to each another!
  4. Mod approved content bots can post up to 10 articles per day.
  5. Threads asking for personal tech support may be deleted.
  6. Politics threads may be removed.
  7. No memes allowed as posts, OK to post as comments.
  8. Only approved bots from the list below, to ask if your bot can be added please contact us.
  9. Check for duplicates before posting, duplicates may be removed

Approved Bots

founded 8 months ago