| Pages in topic: < [1 2 3 4 5 6 7 8 9] > | Consider the consequences of AI before you start a career in translation Thread poster: Gerard de Noord
|
|---|
Sadek_A 
Local time: 17:14
English to Arabic
+ ...
Andreas Baranowski wrote:
Having said that, MT is essentially a good thing, because it excels at some tasks that are horrifying drudgery to do “by hand.” And pre-translation can be immensely helpful also with complex source materials. In the right hands and properly used MT is a wonderful tool which I wouldn’t want to miss.
Every help Machine Translation gets is a step towards a helpless Human Translator. Essentially inevitable in happening, but way far from being essentially low-paid. Doing MTPE jobs is directly perfecting that "tool/aid/etc.", which -in turn- is like carving the really beautiful diva statue from but a mass of clay. A task that is terrifyingly underpriced!
Andreas Baranowski wrote:
I don’t expect translation as a field of work to suddenly disappear but I believe that within a few years practically all translation jobs (IMHO copy is not really translation) will be MTPE jobs, simply because of the enormous cost reduction available at the agency level.
It's a utopian idea for anyone to ever think that the people working on developing MT will continue to make it available for free after it's been perfected.
Andreas Baranowski wrote:
My ideas for Plan B: Before starting in translation, get a degree in a field other than translation that gives you expert knowledge (credibility as a translator) AND can potentially feed you when the going gets tough (my personal choice would be solicitor).
If it can feed, then there is no point in trying translation at all. Outsiders coming into translation are doing so simply because they're unable to be fed by their original fields.
| | | | Baran Keki
Türkiye
Local time: 16:14
Member
English to Turkish
| What about the agencies? | Mar 31, 2022 |
If the MT is so great, won't they get redundant? Why would an end client/company pay an agency 0.20 cents per word if the latter is going to use MT and hire an MT-lover/enthusiast translator to make that translation 'presentable'? Wouldn't they just use the MT and hire that translator themselves instead of dealing with and splashing cash to the agency? So, is that great news for the agencies? It kind of sounds to me like shooting oneself in the foot.
| | | | | Let us eat and drink, for tomorrow we die | Mar 31, 2022 |
Can AI do coding jobs? Everyone wants coders today (and 'first-world' businesses already know that coders from, say, India are better than their compatriots because they are cheaper without being worse), but I'd say AI will become extremely good at writing code before having mastered every aspect of translating hooman languages. I can imagine some AI being provided with all the resources it takes to translate something groundbreaking, for which few if any concepts have ever been expressed in the... See more Can AI do coding jobs? Everyone wants coders today (and 'first-world' businesses already know that coders from, say, India are better than their compatriots because they are cheaper without being worse), but I'd say AI will become extremely good at writing code before having mastered every aspect of translating hooman languages. I can imagine some AI being provided with all the resources it takes to translate something groundbreaking, for which few if any concepts have ever been expressed in the target language, and then doing all the research I would have done and designing a language apparatus suitable for expressing those concepts. But then I'd have to let my imagination go further. Can AI design building projects? Yes. Can it make art? Yes. Can it write marketing texts? Yes. Maybe we should all become entertainers? Would be hard for us translators because we are boring people as someone in this forum said, but hey, AI can entertain too. Maybe we should all strive to excel in sports – like, all of us? But sports are hard, and if we go for cybersports instead, AI will surely beat us. Maybe we should all start doing agriculture to make food for people? Well, one day all our food will be synthesized in a factory by robots and delivered straight to our blood as those of us who remain among the living are busy exploring a VR of their choice. I'm not bothered. I heard something about the skill of being able to put all your winnings together and risking them, and losing, and starting again, and the person describing that skill promised some bonus to all those who have it. Not sure what that bonus was, maybe it's not valuable anymore, but I've been practicing the skill at different times in life just to be safe ▲ Collapse
| | | | | What about the agencies? | Mar 31, 2022 |
Just like a car doesn’t drive by itself but requires a driver to do its job, MT requires a competent operator who is fully familiar with the source and target languages. End clients mostly don’t meet this requirement. The whole point about MT is that it provides a working hypotheses for the final (proper) translation, but nothing more than that. Agencies “sell” this output as “pre-translation” to translators to coerce a massive discount because now its “only” a MTPE job.
| | |
|
|
|
Baran Keki
Türkiye
Local time: 16:14
Member
English to Turkish
| Still though | Mar 31, 2022 |
Andreas Baranowski wrote:
Just like a car doesn’t drive by itself but requires a driver to do its job, MT requires a competent operator who is fully familiar with the source and target languages. End clients mostly don’t meet this requirement. The whole point about MT is that it provides a working hypotheses for the final (proper) translation, but nothing more than that. Agencies “sell” this output as “pre-translation” to translators to coerce a massive discount because now its “only” a MTPE job.
You present sound arguments. But a company that requires regular translation services in various languages could just pick their own translators and assign them PMTE tasks.
Do PMTE tasks get edited/checked by a second translator as is the case with human translations? The so-called 4 eye principle? If all you need is a translator who's happy to do PMTE work for peanuts, then why do you need the middleman (i.e. the agency)?
| | | | Maciek Drobka
Poland
Local time: 15:14
Member (2006)
English to Polish
+ ...
Michael Beijer wrote:
Your post is very relevant to me, since I'm 46 years old and so am planning to be doing this for quite a while still. I have been following the whole AI/MT thing very closely for years, and am hoping it will take a long time before it make us redundant. I've also been trying to figure out a Plan B, in case it all falls apart sooner than I had hoped. Stuff like: learn coding, or some other skill I expect to last longer. I'd be curious to hear about other people's Plan B's...
As a 48-year-old with 20+ years in the translation field, I think I can relate.
While I consider myself a successful professional translator, I am also a natural born pessimist, so I have been considering various plan B's almost ever since I heard the phrase 'Machine Translation'.
After toying with the idea of becoming a coder as well, I decided to dust off my old interest in photography and focus on that as my complementary professional leg. I currently offer commercial pet photography with very positive initial feedback. I am still way too busy with translation to do as much photography as I'd like to, but I intend to keep developing my photography skills, including people portraiture, to achieve a more balanced translation/photography mix in my income stream within a few years.
I am pretty sure at least professional, highly personalised portrait photography is an area where machines won't replace humans in my time.
| | | | | Go for unusual language pairs, learn Ukrainian... | Mar 31, 2022 |
I will not discuss politics here, but from a purely professional point of view, a lot of interpreters are going to be needed in the next few years in face-to-face humanitarian situations, in negotiations, and in industry and reconstruction. Translators will be needed too, and I am not convinced that AI can do it reliably.
A few days ago, President Zelensky held a speech to the Danish Folketing - the country's government and Parliament - and it turned out to be impossible to find an ... See more I will not discuss politics here, but from a purely professional point of view, a lot of interpreters are going to be needed in the next few years in face-to-face humanitarian situations, in negotiations, and in industry and reconstruction. Translators will be needed too, and I am not convinced that AI can do it reliably.
A few days ago, President Zelensky held a speech to the Danish Folketing - the country's government and Parliament - and it turned out to be impossible to find an interpreter who could interpret direct into Danish.
The speech was therefore relayed through English, and members of the Folketing could choose whether they would listen to the English or the Danish version during the 15-minute speech.
Afterwards there was considerable confusion - the two versions had come to different results on important issues in the speech. The administrators excused themselves by pointing out that the EU parliament cannot find enough interpreters from Ukrainian either.
It would probably only have been worse it they had tried to use AI. Zelensky's speech was tailored to the occasion, and the special content required human understanding and knowledge of cultural background and the special situation.
It was not until a day later that correct, written translations were available in both languages - there was still no one who could do it directly from Ukrainian.
It was undoubtedly a challenge, even for the best interpreters, and they have no safety net as we translators have - there is no time to reconsider, go back and check, find a more suitable idiom... I have the deepest respect for interpreters!
The country is talking about receiving 20,000 refugees, perhaps as many as 100,000 ... and we can't train interpreters fast enough!
But as others have said, it is wise to have at least one other specialisation to fall back on.
Firstly because AI will take over most of the basic, generic translation, and you need a specialist area anyway if you are going to succeed at the top end of the market where humans are still needed.
And secondly because you may want to move on after translating for some years.
The profession is changing at an amazing speed, and many of us feel, after more than 20 years in my case, that it is no longer what we signed up for.
I can just retire after the next deadline if it all gets too much, but if you are starting now, you should think seriously about an exit strategy! Language is an art, just like music or sculpture. Perhaps it is a sport, too! Write, teach ... there are lots of opportunities.
There is so much more to learning a language in depth, as a human. With it comes the culture and the history of the country or countries where it is spoken, the thoughts behind the idioms, the literature and poetry, or the climate and geography, the agriculture and industries that have shaped the nation.
I do not believe AI can ever learn all those underlying levels. Humans will always be needed, one way or another. ▲ Collapse
| | | | | I completely disagree | Mar 31, 2022 |
I know I have a long and distinguished track record of playing devil's advocate, but I genuinely and fundamentally disagree with most of what has been said here. The future is bright!
Those starting out now just need to be prepared for a different role where the emphasis is on the latter stages of the translation process: editing, checking, proofreading, reviewing, whatever you want to call it... and less on, well, typing.
Take my translation process:
Step ... See more I know I have a long and distinguished track record of playing devil's advocate, but I genuinely and fundamentally disagree with most of what has been said here. The future is bright!
Those starting out now just need to be prepared for a different role where the emphasis is on the latter stages of the translation process: editing, checking, proofreading, reviewing, whatever you want to call it... and less on, well, typing.
Take my translation process:
Step 1: Dictate 1 hour
Step 2: Polish 4 hours
Step 3: Check 1 hour
MT replaces only Step 1.
Even now, cheapskate agencies would want me to skip Step 2, where I obsess over style and tone and flow. Clients who care will continue to pay for Step 2.
So, MT cuts out 15% of my work. Big deal, I'll take more work or raise my prices or go to the beach.
For translators who don't bother with Step 2, MT cuts out 50% of your work. Not good. Half of you bottom-feeders will have to go elsewhere. But for the other half it will be business as usual.
So I see no reason for MT to drive down any of our earnings. ▲ Collapse
| | |
|
|
|
Tom in London
United Kingdom
Local time: 14:14
Member (2008)
Italian to English
If you could see the badly written academic paper I'm currently translating from Italian to English, you would know there is no AI on earth that would be able to make sense of it. I ran it through MyMemory, Google Translation, and Deepl, and they were all defeated by it.
Try this for size:
ITALIAN
Anche per quanto riguarda il verde, ora selvatico e infestante, viene proposto un intervento puntuale, caso per caso, tuttavia corrispondente al nuovo disegno d... See more If you could see the badly written academic paper I'm currently translating from Italian to English, you would know there is no AI on earth that would be able to make sense of it. I ran it through MyMemory, Google Translation, and Deepl, and they were all defeated by it.
Try this for size:
ITALIAN
Anche per quanto riguarda il verde, ora selvatico e infestante, viene proposto un intervento puntuale, caso per caso, tuttavia corrispondente al nuovo disegno della Cittadella.
Per quanto concerne il restyling, la rifunzionalizzazione degli edifici e le nuove architetture, le proposte emerse dal lavoro dei gruppi sono, in sintesi, le seguenti
ENGLISH (DEEPL)
As for the greenery, which is now wild and weedy, a specific intervention is proposed, case by case, but corresponding to the Citadel's new design.
With regard to the restyling, refurbishment of the buildings and new architecture, the proposals that emerged from the groups' work are, in brief, as follows
I rest my case. Translating a text like this begins with rewriting the original, paragraph by paragraph, in the source language.
The market for translation is still open for experts who are willing to put in the work (and charge the full rate).
[Edited at 2022-03-31 16:03 GMT] ▲ Collapse
| | | | Tom in London
United Kingdom
Local time: 14:14
Member (2008)
Italian to English
Ice Scream wrote:
MT replaces only Step 1.
And in many cases not even that. See my previous post.
| | | | Sadek_A
Local time: 17:14
English to Arabic
+ ...
We're unfortunately, once again, going in circles.
Let's pinpoint, so that the discussion can provide something useful:
1. Algo/AI/Machine/etc. is an EXISTING technology being USED to REPLACE us. We can't eliminate it from existence, but we can do something else, as will be listed below.
2. Any text, general or specialized, simple or complex, well- or horribly-written, can be run through Algo/AI/Machine/etc. by agency/company/client/etc. and offered to tran... See more We're unfortunately, once again, going in circles.
Let's pinpoint, so that the discussion can provide something useful:
1. Algo/AI/Machine/etc. is an EXISTING technology being USED to REPLACE us. We can't eliminate it from existence, but we can do something else, as will be listed below.
2. Any text, general or specialized, simple or complex, well- or horribly-written, can be run through Algo/AI/Machine/etc. by agency/company/client/etc. and offered to translator/linguist/etc. as MTPE job.
3. When you do the translation all on your own, your target text will be used -with or without, but mostly without, your consent- to teach Algo/AI/Machine/etc. (a LOWER stage)
4. When you do MTPE job, your target text is actually used to perfect what the Algo/AI/Machine/etc. has been taught. (a HIGHER stage)
5. MTPE jobs are low-priced under the pretense that most of the job was done by the Algo/AI/Machine/etc.
But, it wasn't, was it? Algo/AI/Machine/etc. stole your work and the work of others, and now it needs you to launder that stolen corpus. Does this task look like something that can be low-priced?
6. Once the Algo/AI/Machine/etc. is perfected and you -human translator- are no longer needed, owners of that new intellectual property (that actually was but your intellect and your property!) will be swimming in cash.
Solution:
Your MTPE rate must be higher than your translation rate, and your translation rate must be high enough to begin with, because the help you provide today will be the reason you're jobless tomorrow!
Tom in London wrote:
put in the work (and charge the full rate).
✓✓✓ ▲ Collapse
| | | | Dan Lucas
United Kingdom
Local time: 14:14
Member (2014)
Japanese to English
| Gotta wear shades | Mar 31, 2022 |
Ice Scream wrote:
So I see no reason for MT to drive down any of our earnings.
Agreed. Given the number of unfalsifiable assertions already floating about on this thread, I don't have any qualms about making one of my own: things are actually looking pretty good.
Sure, CAT and MT abstract away the boilerplate, but there's still a need for people who can give nuanced and thoughtful translations, and explain their choices. As Chris points out, Step 1 is the easy bit. As for Step 2, I'm a chatty translator whose natural instinct is to leave a paper trail of comments, alternative phrasings, examples of usage, advice, and warnings. This is time-consuming, and I don't necessarily expect them to be read, but they are there for those who are interested. The agencies regularly feed back favourable comments by end clients who like this approach.
As for business trends, my segment of the market had a bumper year in 2021. I saw a steady increase in the number of approaches by a variety of clients, all wanting to lock me in for more work. Here at the end of the first quarter of 2022, there's a continued sense of constrained supply and very strong demand, which suggests that this isn't a reflection of any talents I might have, so much as a reflection of conditions in the sector.
Multiple agencies have already warned me of an abnormally busy season ahead in late April/May. I have nearly 30 jobs booked in the schedule, some having gone in as early as late February. "There are many more to come" according to the rather menacing words of one client. I can't believe it will be busier than May 2021, which was a record for me (and probably others) but that's what they're saying. We'll see. Talk is cheap. It might not even get to the levels of last May. I hope not. I was very tired by the start of June 2021.
My sense is that this is a cyclical issue caused by a confluence of several different factors (regulatory changes, restructuring of the Tokyo market, backlog from COVID, etc.), which has left end clients with a pile of financial documentation to get translated and out of the way. Later in the year demand will probably slow sharply, so don't spend it all at once, boys and girls!
Then again, every 2-3 years we seem to get this kind of bulge in demand, MT or no MT, CAT or no CAT. The constituent factors that drive it may vary, but something always turns up. After all, the defining feature of a cycle is that it recurs.
Dan
| | |
|
|
|
| Macro trends/MT impact and limits/Barrier to entry | Mar 31, 2022 |
MACRO TRENDS
Since we're discussing macro trends, is the industry shrinking or expanding, on a yearly basis? Is less content being translated, globally, with each passing year? Or is it quite the contrary? And while we are at it, how does the translation market fair when compared to the overall work/economy situation? Is it completely isolated from everything else?
Is MT the only factor going on? Let's beware of reductionism.
MT IMPACT AND LIMITS
... See more MACRO TRENDS
Since we're discussing macro trends, is the industry shrinking or expanding, on a yearly basis? Is less content being translated, globally, with each passing year? Or is it quite the contrary? And while we are at it, how does the translation market fair when compared to the overall work/economy situation? Is it completely isolated from everything else?
Is MT the only factor going on? Let's beware of reductionism.
MT IMPACT AND LIMITS
MT does has a more or less significant impact (and it is humbling to see some established and esteemed colleagues express their worry about the future of the profession in such ways), and it would be advisable to take an informed stance, and be aware of its usefulness and limits, just as we ought to know the usefulness and limits of bilingual dictionaries (translation equivalence is established at the discourse [textual] and not at the language level [on which dictionaries operate], and translators do not translate words; they translate their sense in a given context).
BARRIER TO ENTRY
Maybe MT is the barrier to entry that so many serious practitioners have been dreaming and clamoring about for so long.
At last a chance to weed out the dilettantes!
Why don't I hear hurrays?
[Edited at 2022-03-31 17:03 GMT] ▲ Collapse
| | | |
Hey Dan, it is so great to see you around!
| | | | Metin Demirel
Türkiye
Local time: 16:14
Member (2018)
Italian to Turkish
+ ...
| MT and English hand-in-hand | Mar 31, 2022 |
Christine Andersen wrote:
Parliament - and it turned out to be impossible to find an interpreter who could interpret direct into Danish.
The speech was therefore relayed through English, and members of the Folketing could choose whether they would listen to the English or the Danish version during the 15-minute speech.
The threat, for us translators, is not only the advance of technology, but also the increasing prevalence of the English language. As far as I know, MT uses the English language as the middle man when it translates between two other languages. For example, the Danish translation of a Ukrainian text by MT is actually the EN>DA machine translation from the output of UA>EN machine translation, which is not shown to the user.
I once received a so-called source text in English that contained inconsistent sentences like,
"Control that...
Check that...
Checking that....
Verify that... " etc.
I know the real source text in Italian was "Verifica che..." but the company (and/or the agency) had resorted to basing their entire translation project on a low-quality English text created by MT from Italian.
Now the non-English speaking companies employ English copyrighters, or they prefer to have their material translated from their native language into English first, to be then translated into other languages, instead of translating them directly from the company's native language, thus rendering UA>DA translators merely, for lack of a better word, useless.
[Edited at 2022-03-31 17:05 GMT]
| | | | | Pages in topic: < [1 2 3 4 5 6 7 8 9] > | There is no moderator assigned specifically to this forum. To report site rules violations or get help, please contact site staff » Consider the consequences of AI before you start a career in translation | Protemos translation business management system | Create your account in minutes, and start working! 3-month trial for agencies, and free for freelancers!
The system lets you keep client/vendor database, with contacts and rates, manage projects and assign jobs to vendors, issue invoices, track payments, store and manage project files, generate business reports on turnover profit per client/manager etc.
More info » |
| | 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! » |
|
| | | | X Sign in to your ProZ.com account... | | | | | |
Login to reply/comment 
Translation news
