Pages in topic:   [1 2] >
Poll: What is your opinion on the advances in neural machine translation?
Thread poster: ProZ.com Staff
ProZ.com Staff
ProZ.com Staff
SITE STAFF
Mar 21, 2017

This forum topic is for the discussion of the poll question "What is your opinion on the advances in neural machine translation?".

View the poll results »



 
Julian Holmes
Julian Holmes  Identity Verified
Japan
Local time: 07:35
Japanese to English
My opinion? Mar 21, 2017

More like 'neurotic' machine translation

I really don't think MT can be applied to my language pair. It's nothing to get neurotic about.


 
neilmac
neilmac
Spain
Local time: 23:35
Spanish to English
+ ...
Mixed feelings Mar 21, 2017

Luckily I'm old enough to be in a situation where I will probably be able to make a living from translating until I run out of steam. However, it does seem as though the much vaunted increased efficiency of neural MT will lead to post-editing skills becoming more of a requirement, with less emphasis on translation per se. Then again, there are areas of translation which lend themselves more to MT than others. I don't think computers will be writing great literature any time soon.

 
Chié_JP
Chié_JP  Identity Verified
Japan
Local time: 07:35
Member (2013)
English to Japanese
+ ...
Hope for DTP Mar 21, 2017

Is Neural DTP ready yet?
Even Word can not fix the size and placement of tables and diagrams so it helps us to use something that takes care of lengthy tags. So fluctuating PPT cursors...Excel pointer goes wrong...do they learn at all or are they willing to?

One issue with neural is security. Once "learned by machine" you can not erase it because even the developers say they do not know what and how they learn the big data.

I hear machines are better and fast at
... See more
Is Neural DTP ready yet?
Even Word can not fix the size and placement of tables and diagrams so it helps us to use something that takes care of lengthy tags. So fluctuating PPT cursors...Excel pointer goes wrong...do they learn at all or are they willing to?

One issue with neural is security. Once "learned by machine" you can not erase it because even the developers say they do not know what and how they learn the big data.

I hear machines are better and fast at writing from scratch than at translating,
just like humans are! Probably they are not good at formatting either?
Then I do hope they can help us at proofreading job (pointing out unusual row of characters, grammar check etc) after translation
- they are less rewarding tasks so it would be nicer world if machines can take care of it all.

[2017-03-21 08:44 GMTに編集されました]

[2017-03-21 08:45 GMTに編集されました]
Collapse


 
Maria Silvestri
Maria Silvestri  Identity Verified
United States
Member (2010)
Italian to English
+ ...
Things are going to change. Mar 21, 2017

I agree with neilmac.

It's better to be aware of progress in AI translation and ready to change with the times (being prepared to shift more toward post-editing, for example) than to be in denial about it and say that it's impossible.

The technology is still in its infancy, (for example, I am not seeing many post-editing jobs in my pairs yet, for example) but I am following news of it closely and I know that if I want to stay hireable as a freelancer in this industry, I
... See more
I agree with neilmac.

It's better to be aware of progress in AI translation and ready to change with the times (being prepared to shift more toward post-editing, for example) than to be in denial about it and say that it's impossible.

The technology is still in its infancy, (for example, I am not seeing many post-editing jobs in my pairs yet, for example) but I am following news of it closely and I know that if I want to stay hireable as a freelancer in this industry, I need to be agile and aware.

I think we're 10-20 years out from some major changes to the way we do things because of the advances being made.
Collapse


 
Nikki Scott-Despaigne
Nikki Scott-Despaigne  Identity Verified
Local time: 23:35
French to English
Not ready yet Mar 21, 2017

It should be an improvement on machine translation based on statistical models, but we're a long way from being able to put any old text into a translation sausage-making machine and producing a text that is digestible.

 
Francisco ABREU
Francisco ABREU  Identity Verified
Brazil
Local time: 19:35
Member (2005)
English to Portuguese
+ ...
Obsolete Mar 21, 2017

It's obvious to anyone who understands even a little about technology that advancing algorithms will render our profession obsolete in the future. How soon it's difficult to predict, but maybe even in our lifetime.

Sooner or later, translators, like a large number of other professionals, will become history.

Men will overcome the Tower of Babel episode and live in a world without language barriers. How that world will shape itself remains to be seen – but it's inevita
... See more
It's obvious to anyone who understands even a little about technology that advancing algorithms will render our profession obsolete in the future. How soon it's difficult to predict, but maybe even in our lifetime.

Sooner or later, translators, like a large number of other professionals, will become history.

Men will overcome the Tower of Babel episode and live in a world without language barriers. How that world will shape itself remains to be seen – but it's inevitable.
Collapse


 
No opinion Mar 21, 2017

What I want to know is how the 50% expressing an opinion here have acquired the knowledge needed to form an opinion on this subject.

I, for one, don't have a clue. Unless you're not a translator but a neural MT developer, how would you know?


 
Jan Truper
Jan Truper  Identity Verified
Germany
Local time: 23:35
English to German
Hooked on Phonics worked for me! Mar 21, 2017

Chris S wrote:

What I want to know is how the 50% expressing an opinion here have acquired the knowledge needed to form an opinion on this subject.



Hooked on Phonics worked for me!


This exchange would be one example of why I personally am not worried about AI, as far as my profession as a translator is concerned. I doubt there'll ever be a computer that is able to connect the dots between the question and a pesky 90s US TV commercial to come up with a useful translation (basically, in this context "Hooked on Phonics worked for me!" is a slightly offensive way of saying "by reading" -- no offense meant, Chris).

I translate mostly dialogue. Until there's a computer that can take humor, cultural references, facial expressions, tone and emotional context into account when translating, I feel safe.
And if there's ever such a computer, I hope it'll also be able to cook and like romantic walks on the beach


 
Noni Gilbert Riley
Noni Gilbert Riley
Spain
Local time: 23:35
Spanish to English
+ ...
Head firmly buried in sand... Mar 21, 2017

...I can be heard to mutter "thank goodness I'm 57!".

 
Judith Shiozawa (X)
Judith Shiozawa (X)  Identity Verified
Germany
Local time: 23:35
German to English
Neural machine translation2 Mar 21, 2017

I agree with neilmac. My active translation career will probably come to a close before these new translation systems gain a competitive edge over human translators. Honing one's editing skills seems to be sound advice for those intending to work for a longer time in the industry.

 
Maria Teresa Borges de Almeida
Maria Teresa Borges de Almeida  Identity Verified
Portugal
Local time: 22:35
Member (2007)
English to Portuguese
+ ...
Other! Mar 21, 2017

I have no crystal ball and I won’t probably be around when these "advances" happen, but let’s not forget that my father worked until the day he died (at 93) and my mother (she is 103) says that the worst decision she ever took was to stop working at 95…

Crystal_ball


 
Mario Freitas
Mario Freitas  Identity Verified
Brazil
Local time: 19:35
Member (2014)
English to Portuguese
+ ...
It all depends on us Mar 21, 2017

MTs are evolivng, no doubt. But there is no chance they'll ever replace all translators in the future. MTs are very useful in this globalized world, and indeed the Babel Tower is history. But professional translaton is something else. There is no mathematical formula for language, no algorithm able to make a machine interpret messages, moods, humor, feelings, typos, poor construction, Chinglish, etc. The best possible algorithms will allow the machine to translate text only, with none of these f... See more
MTs are evolivng, no doubt. But there is no chance they'll ever replace all translators in the future. MTs are very useful in this globalized world, and indeed the Babel Tower is history. But professional translaton is something else. There is no mathematical formula for language, no algorithm able to make a machine interpret messages, moods, humor, feelings, typos, poor construction, Chinglish, etc. The best possible algorithms will allow the machine to translate text only, with none of these factors.

So, yes, a good part of the translators will have to find another job, because part of the translations involve text only. But the quality professionals and translations will still exist for many generations.

Now one thing we should all do, as professionals, is to refuse vehemently any type of the so-called "post-editing" jobs for a lower price than the translation. Those who do that are contributing for a translation world dominated by machines.

[Edited at 2017-03-21 15:34 GMT]
Collapse


 
Trevor Chichester
Trevor Chichester  Identity Verified
United States
Local time: 17:35
Member (2012)
German to English
+ ...
The future depends largely on US Mar 21, 2017

What a lot of translators somehow fail to realize is that the "future" of MT is largely dependent on US.

If we don't want our professions going the way of the dodo we must ALL absolutely, 110% REFUSE to work on anything MT related...ESPECIALLY that "post-edit" garbage that some companies somehow market to clients as "professional, reliable quality".

This is going to play out exactly how the drastic drop in rates happened.....WE LET IT HAPPEN. Instead of all of us vehem
... See more
What a lot of translators somehow fail to realize is that the "future" of MT is largely dependent on US.

If we don't want our professions going the way of the dodo we must ALL absolutely, 110% REFUSE to work on anything MT related...ESPECIALLY that "post-edit" garbage that some companies somehow market to clients as "professional, reliable quality".

This is going to play out exactly how the drastic drop in rates happened.....WE LET IT HAPPEN. Instead of all of us vehemently opposing ridiculous, laughably impossible deadlines accompanied by outrageously low rates, some of us continued to work under these conditions and now look what's happened. Rates have dropped to an all time low because not all of us had a backbone to stand up and say "NO, we won't work for these rates and under these conditions".

The future of MT is determined by how we as a community react. We must all unequivocally OPPOSE any kind of MT work. If we don't stand as a community our professions will absolutely cease to exist.


[Edited at 2017-03-21 16:08 GMT]

[Edited at 2017-03-21 17:24 GMT]
Collapse


 
DZiW (X)
DZiW (X)
Ukraine
English to Russian
+ ...
weird copy-booking Mar 21, 2017

Unlike a mere statistical approach, everything that can be measured, fuzzy-weighted, and described via rules and exceptions is a subject to deep learning in neural "connectionist" networks, which almost imitate the human axons... For almost 40 years.

It's not just a matter of time for besides mostly supervised data mining process it has rather many problems with assessing/classification patterns, which often are not such even in a stereotyped human world; not a cut-and-dried solutio
... See more
Unlike a mere statistical approach, everything that can be measured, fuzzy-weighted, and described via rules and exceptions is a subject to deep learning in neural "connectionist" networks, which almost imitate the human axons... For almost 40 years.

It's not just a matter of time for besides mostly supervised data mining process it has rather many problems with assessing/classification patterns, which often are not such even in a stereotyped human world; not a cut-and-dried solution.

Nowadays there're two strategies: either improve a machine, or improve a human--and both are mostly but a marketing actions.
Collapse


 
Pages in topic:   [1 2] >


To report site rules violations or get help, contact a site moderator:

Moderator(s) of this forum
Jared Tabor[Call to this topic]

You can also contact site staff by submitting a support request »

Poll: What is your opinion on the advances in neural machine translation?






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! »
Trados Studio 2022 Freelance
The leading translation software used by over 270,000 translators.

Designed with your feedback in mind, Trados Studio 2022 delivers an unrivalled, powerful desktop and cloud solution, empowering you to work in the most efficient and cost-effective way.

More info »