Poll: Have you noticed any improvement in MT quality in the past 12 months?
Thread poster: ProZ.com Staff
ProZ.com Staff
ProZ.com Staff
SITE STAFF
Mar 15, 2025

This forum topic is for the discussion of the poll question "Have you noticed any improvement in MT quality in the past 12 months?".

View the poll results »



 
Alan Corbo, CT
Alan Corbo, CT  Identity Verified
Uruguay
Local time: 12:14
English to Spanish
+ ...
Not really Mar 16, 2025

I guess we should first agree on a quality benchmark for translation. What is a "good" translation? One that is not too literal? Sometimes translations need to be literal to convey the message faithfully. But other times you need to do exactly the opposite (i.e., avoid literality at all costs) for the translation to be "good". And don't get me started with considerations about target... See more
I guess we should first agree on a quality benchmark for translation. What is a "good" translation? One that is not too literal? Sometimes translations need to be literal to convey the message faithfully. But other times you need to do exactly the opposite (i.e., avoid literality at all costs) for the translation to be "good". And don't get me started with considerations about target audience, cultural nuances, context of reception, historical aspects, linguistic economy and theme and rheme, to name just a few dimensions translators need to be constantly mindful of as they progress through their tasks.

I don't use MT in my daily work, but the few times I've had to edit/revise work performed using, let's say, DeepL, the "machine" part of the process is glaringly obvious. It shows through in all its glory. Mistakes made by humans are not the mistakes made by machines. Machines usually follow the sequence of words in a sentence almost to the letter (with a few exceptions, like converting the English passive voice to active voice in Spanish), and that results in rather stilted, unidiomatic translations, or so I've seen. That hasn't really changed much since the introduction of neural machine translation, around 2017. As for AI, when you ask ChatGPT to give more "flair" to a translation, it very quickly starts adding things that were not there to begin with, in a sort of hallucinatory process that embelishes that final product, but that is far from being faithful to the original. Very dangerous stuff if left unchecked.

What MT/AI has definitelly managed to do is to standardise what an "ok" translation is, and given its lack of creativity (a language is much more than a system of patterns, but machines deal with them as such), I'd dare say translations (and language in general) are bound to become mass-produced, undistinguisable pieces of commodity. After all, do you think the likes of Homer, Virgil, Cervantes or Shakespeare would've been able to do what they did if they had been "helped along" by ChatGPT to generate ideas for their work? What a loss that would've been.

[Edited at 2025-03-16 05:13 GMT]
Collapse


Maria Laura Curzi
Daryo
Liena V.
 
Johan Beyens (X)
Johan Beyens (X)
Belgium
Local time: 17:14
French to Dutch
+ ...
how about 12 years Mar 16, 2025

12 months? I haven't noticed much of an improvement in the past 12 years.

Lingua 5B
Maria Laura Curzi
Lija Lija
Daryo
Liena V.
 
Evgeny Sidorenko
Evgeny Sidorenko
Russian Federation
Local time: 18:14
English to Russian
+ ...
Degradation? Mar 16, 2025

In the last approximately 6-9 months I have noticed strange bugs in the engine that I use (i.e. absolutely unacceptable, unexpectedly strange, randomly occurring translations of some source text/words) for no apparent reason. Not often, but I have not noticed this kind of thing before.

 


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: Have you noticed any improvement in MT quality in the past 12 months?






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! »
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 »