Poll: Is medical/legal linguistic specialisation still a good niche in EU? Initiator des Themas: ProZ.com Staff
|
|---|
This forum topic is for the discussion of the poll question "Is medical/legal linguistic specialisation still a good niche in EU?".
This poll was originally submitted by Edita Szcsukova. View the poll results »
| | | | | Both, I would think | Feb 1, 2024 |
Any specialisation is a good niche anywhere, by definition.
Both of these are areas where MT will have issues understanding the source because they tend to be badly written and/or complex. On the other hand, both require pretty literal translations, which would favour MT.
| | | | Dan Lucas 
Vereinigtes Königreich
Local time: 04:38
Mitglied (2014)
Japanisch > Englisch
| An old favourite | Feb 1, 2024 |
Christopher Schröder wrote:
because they tend to be badly written and/or complex.
| | | | | There is going to be bread for sworn translators... | Feb 1, 2024 |
for as long as there are bureaucrats needing to be fed a steady diet of documents, all duly stamped and inscribed with the magic runes: "I hereby certify the above is a true and correct translation of the original document".
Even if it's going to get all digital at some point, we're probably looking at decades worth of paper that still needs to be examined by hand and have all the stamps and watermarks and whatnots accounted for in translator's notes so that it can be used officiall... See more for as long as there are bureaucrats needing to be fed a steady diet of documents, all duly stamped and inscribed with the magic runes: "I hereby certify the above is a true and correct translation of the original document".
Even if it's going to get all digital at some point, we're probably looking at decades worth of paper that still needs to be examined by hand and have all the stamps and watermarks and whatnots accounted for in translator's notes so that it can be used officially.
Just keep your fingers crossed and trust in the bureaucracy to expand as it does to meet the needs of the expanding bureaucracy.
Same could be said for scribble with no hope for any sort of ocr, but I see now that this jab has already been thrown.
Still, having worked on some fine specimens (treatment notes: mostly, but thankfully not all handwritten), it's not something I could be easily persuaded to do on a daily basis. ▲ Collapse
| | |
|
|
|
Alex Lichanow
Deutschland
Local time: 05:38
Englisch > Deutsch
+ ...
While I am not really getting a good look at this through my crystal ball, my personal observation is that at least DeepL is frighteningly good at translating legalese, at least from English to German and vice versa. Might be the weird, unnatural wording that makes legal an ideal candidate for AI.
| | | | Anton Konashenok
Tschechische Republik
Local time: 05:38
Französisch > Englisch
+ ...
| Both will remain good for the foreseeable future | Feb 2, 2024 |
Alex Lichanow wrote:
While I am not really getting a good look at this through my crystal ball, my personal observation is that at least DeepL is frighteningly good at translating legalese, at least from English to German and vice versa. Might be the weird, unnatural wording that makes legal an ideal candidate for AI.
My observation is exactly opposite: automatic systems produce language that is mostly usable, but every once in a while they make errors so gross that trusting them becomes dangerous, especially in subject fields like law and medicine. As the AI is inherently unable to understand the text, I see no pathway to improvement until AGI becomes reality.
Christopher Schröder wrote:
On the other hand, both require pretty literal translations, which would favour MT.
Only partially so. Both occasionally contain highly convoluted phrases that can easily confuse both MT and the human reader. A good translator will preserve the meaning but make the phrase more readable. Furthermore, in both law and medicine, individual countries often have idiosyncratic traditional concepts that are either commonly misunderstood or even absent altogether in other countries, and should never be translated literally. One example is "altération de l'état général" in French, which I've seen innumerable times being translated into English as "general deterioration" but should properly be translated as "wasting syndrome" or "cachexia".
[Edited at 2024-02-02 07:36 GMT]
| | | | Lieven Malaise
Belgien
Local time: 05:38
Mitglied (2020)
Französisch > Niederländisch
+ ...
Anton Konashenok wrote:
automatic systems produce language that is mostly usable, but every once in a while they make errors so gross that trusting them becomes dangerous, especially in subject fields like law and medicine. As the AI is inherently unable to understand the text, I see no pathway to improvement until AGI becomes reality.
Agreed. Many translators seem to think that MT will continue to improve eternally (and I understand that, we are overwhelmed with 'good news' about AI), but that's just not possible unless, as you say, machines will equal human intelligence and feel like humans do. MT can be good, but 'makes' mistakes at a regular basis and that simply won't change. So, at least human intervention will continue to be necessary.
I also hear people say that MT produces output of which at least 80% doesn't have to be modified. I'm sorry, but I'm sure that says more about the quality standards of those people than about the genuine quality of MT. Again, MT can be surprisingly good, but it delivers far from finished quality products.
| | | | Alex Lichanow
Deutschland
Local time: 05:38
Englisch > Deutsch
+ ...
Lieven Malaise wrote:
I also hear people say that MT produces output of which at least 80% doesn't have to be modified. I'm sorry, but I'm sure that says more about the quality standards of those people than about the genuine quality of MT. Again, MT can be surprisingly good, but it delivers far from finished quality products.
In fact, AI translators are caught in a loop of self-propagating garbage, as Amazon recently noticed:
https://slator.com/amazon-flags-problem-using-web-scraped-machine-translated-data-llm-training/
(It's not like I have been preaching this for years, but hey, I'm just a small freelance translator with no connection to our new artificial gods.)
I still have some hope that all the cr*p-tier AI/machine translations of lately are going to come crashing down hard on those that trust them blindly.
[Edited at 2024-02-02 08:13 GMT]
| | |
|
|
|
neilmac
Spanien
Local time: 05:38
Spanisch > Englisch
+ ...
At least, I would hope so. But then again, I'm from the 20th century
| | | | Anton Konashenok
Tschechische Republik
Local time: 05:38
Französisch > Englisch
+ ...
Lieven Malaise wrote:
I also hear people say that MT produces output of which at least 80% doesn't have to be modified. I'm sorry, but I'm sure that says more about the quality standards of those people than about the genuine quality of MT.
Exactly, and by the way, translation agencies' typical threshold for hiring human translators is the ability to produce translations where 98-99% doesn't have to be modified by the editor! 80% looks worthy of instant disqualification for professional incompetence and lack of diligence.
| | | | To report site rules violations or get help, contact a site moderator: You can also contact site staff by submitting a support request » Poll: Is medical/legal linguistic specialisation still a good niche in EU? | 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! » |
| | Pastey | Your smart companion app
Pastey is an innovative desktop application that bridges the gap between human expertise and artificial intelligence. With intuitive keyboard shortcuts, Pastey transforms your source text into AI-powered draft translations.
Find out more » |
|
| | | | X Sign in to your ProZ.com account... | | | | | |
Login to reply/comment 
