| Pages in topic: < [1 2 3 4] | Poll: Would you be willing to work on a project that contributes to development of AI translation? Thread poster: ProZ.com Staff
|
|---|
| That's because it was "temp" | Mar 7 |
Christopher Schröder wrote:
I thought it would work like that but both times I’ve mistakenly accepted jobs that have been run through MT, matches have been taken from the TM and the MT has then translated new text without any reference to what was in the TM. There was no learning at all.
That was agency-owned MT, and the text had been pre-populated without confirmed entries. Your entries will be entered in the "corpus" (TM) after it's finalized. Agency-owned MT is really bad in the first few texts because the TM is largely empty, but since it's usually client-specific or industry-specific, it gets really good really fast (within only 8-10 similar projects).
If you do it on Google Trans, where the subsequent cells are not populated yet, and you keep a steady time pace from one cell to the next, you'll see your "unique" terms starting to appear by themselves after 3-4 entries or so. You'll still have to fix quite a few expressions, because Google-trans has a huge non-specific messy TM (so it's not as good as agency-MT even after hundreds of projects). But you'll agree that the mildly edited result is "good enough for the price" for the majority of generic texts. Especially when the original says "Translations of this text are for easy reference only, and for all legal matters the original source text prevails".
| | | | | Yes, publicly available MT does that | Mar 7 |
Liena Vijupe wrote:
I have seen the results changing sometimes, but it's not very different from a new TM match simply replacing a previous one, it is not learning anything.
Yes, publicly available generic MT does that. It switches back and forth, depending on the number of matches in its TM.
I believe it has reached the peak already and will only go down from now with the input quality worsening.
Only if the input by the translators gets worse. And even if it does, that will not be an argument sufficient for agencies to pay more, when there are thousands who would accept the project at even lower prices than the initial quoted. Remember that your price for MTPE is the same as the price for straight translation for thousands of your colleagues.
At any rate, the only defense I see for translators is a second (and/or third) source of income, and let the chips fall where they may in the translation industry.
| | | | | "Never perfect" doesn't matter | Mar 7 |
Liena Vijupe wrote:
I think it shows the limits of MT, it can never be perfect because it does not understand the message. The whole idea of "learning" implies it is being fed better and better translations of the same thing and is also able to select the best and most appropriate among millions of matches each time which is neither true, nor realistic.
It doesn't matter. What matters is what we get paid per unit. On an average 5,000-word text, 15 years ago, we were paid $0.12 + formatting $30/hour (so usually $630). Inflation-adjusted it should have been $700 at least. But today it's only about $180 (and we also have the fuzzies/matches). And the volume won't increase (only sporadically), because a translation is not a consumer product (people buy another pair of shoes when they have money, but not more translations).
Now, try to convince end-clients and project managers that your straight text is so much better than edited MT to justify the price increase... thousands of Western companies bankrupted from cheaper competition even when the quality difference was obvious to everyone, because the issue of cost is huge. Translators do have more room to move, theoretically, because they could just ask for higher prices from the agencies (the large ones work with scandalously high markups). But with tens of thousands of translators trying to get the projects at lower prices each year, I don't see this ever happening.
Anyone with the slightest understanding of software development would concur this is utter nonsense.
For software development in the sense of total construction of the entire software suite, yes, you're right. AI can't write even a simple image display program from beginning to the end. But much of the "subroutine coding only" part has already been replaced, even before AI. Unfortunately, in the translation industry, translators are not doing the "whole document process construction". We' re only the "coding" part.
| | | |
Tom in London wrote: Abdulsamad Attayee wrote:
I think MTPE might already be assisting with the development of AI translation... Yes, we are probably all collaborating without actually knowing about it. Just using Deepl probably implicates us. But there's a difference between collaborating because it's not possible not to, and collaborating intentionally and enthusiastically in one's own demise.
PS) Something went wrong I think. The image says 'Linguee + AI = DeepL Translator' (just in case you can't open the image)
[Edited at 2024-03-07 12:38 GMT]
[Edited at 2024-03-07 12:38 GMT]
| | | | | Pages in topic: < [1 2 3 4] | To report site rules violations or get help, contact a site moderator: You can also contact site staff by submitting a support request » Poll: Would you be willing to work on a project that contributes to development of AI translation? | CafeTran Espresso | You've never met a CAT tool this clever!
Translate faster & easier, using a sophisticated CAT tool built by a translator / developer.
Accept jobs from clients who use Trados, MemoQ, Wordfast & major CAT tools.
Download and start using CafeTran Espresso -- for free
Buy now! » |
| | Wordfast Pro | Translation Memory Software for Any Platform
Exclusive discount for ProZ.com users!
Save over 13% when purchasing Wordfast Pro through ProZ.com. Wordfast is the world's #1 provider of platform-independent Translation Memory software. Consistently ranked the most user-friendly and highest value
Buy now! » |
|
| | | | X Sign in to your ProZ.com account... | | | | | |
Login to reply/comment 
