Trends in the language services industry: Artificial intelligence

Artificial intelligence. Forget whether it's really intelligent, it's here. What is it doing?
 
Quotes from language professionals:

An interpreting contract was cancelled because AI was implemented to LIVE translate & subtitle a news show where I was working.
 
Technological advances are not only exciting, but greatly impact the development of our industry. We need to adapt and transform ourselves, since technology is dynamic as well.
 
I am not extremely worried about technological advances, on the contrary I am trying to specialise so that I can make them work to my advantage.
 

AI can write, AI can translate

 
There has been much talk in recent years about Artificial Intelligence (AI) and how “disruptive” it is or will be. Most translators and interpreters agree that AI is already affecting their world and has good, practical uses in the industry, but at the moment most are hard pressed to find clear examples of this in their daily work, outside of its use in conjunction with machine translation.
 
The world, and the industry, are seeing an increase in the automation of communication, security, projects, and workflows, thanks to the assistance of AI. Machine learning and artificial intelligence are shaping how and what we translate. When we talk about the increasing creation of content we should keep in mind that AI is out there doing its fair share of content creation now, and in some cases it is performing the role of both content creator and translator. With varying degrees of success, of course. 
 
Quotes from language professionals:

The impact that it has (other than the enquiries not being placed with a real translator) is that agencies appear to make more use of software, including AI, to judge a translator's work, assessing it, then, in a mechanical fashion, thus often incorrectly.
 
 
And as a communicator in that automation of workflows, AI is only as effective as we make it. One example which is discussed frequently among translators: an email which is actually part of an automatic set of emails fired off by a translation management system (TMS) searching for available providers is taken by the translator to have originated from a hurried, culturally-insensitive human project manager when certain niceties such as the recipient’s name or a greeting or sign off or other data are excluded. The effects of this sort of thing, which tends to rub professionals who deal very closely with language the wrong way, are relatively easy to avoid, whether you are a human PM or a human setting up your TMS to sound like a human PM. But the devil is in the details, right?
 

A human-machine-human loop

Quotes from language professionals:

It has a bad influence on style and choice of terminology.
 
Translators report an interesting phenomenon which is at least algorithm-based if not purely an effect of AI: some clients are now requiring that target terms in translations be terms which have the highest hit rates in search engines, even if those terms are not the best match for the text itself being translated. In this loop, we’ve told the machine what to do, the machine shows us that in order to get the highest rate of attention to a service or product we should write in a given way, and then translate in a given way, altering the way we communicate.  
 
Quotes from language professionals:

Intelligence agencies use AI to scan the vast body of data that they need to follow. In the old days, translators were used to translate messages, technical articles, etc., for these agencies, but today there aren't enough translators, and there isn't enough money to pay them, to translate all the material that needs to be scanned. So in that sense, AI has affected the market. But the quantum increase in data has changed the market as well, forcing the need to develop AI.

On the other hand, between the ease of generating text with word processing, the means to circulate it on the Internet, and ever-more contact between countries, the world is increasingly flooded with words that are eligible for translation.

https://www.proz.com/post/2808386#2808386
 
 
Those lucky interpreters don’t have to worry about any of this, right? Well, instant subtitling or closed captioning technology is getting better and better, and might be an adequate solution in some cases. Remote interpreting relies heavily on systems which can know who is online and available in order to route calls quickly and efficiently, and can learn who to route to first based on response times, call frequency, and duration.
 
Quotes from language professionals:

Potent search machines like Google condition more and more the choice of adequate terms. For example, some clients explicitly require the use of terms that have the highest rate of hits in Google Analytics even if they are not the best for describing and selling their products.
 

It could go either way. What I believe I am seeing as an effect of better AI tools in translation, is an increase in the number of jobs and shorter work cycles. And perhaps a tendency to lower tariffs for menial tasks. On the other hand, high quality, highly specialized tasks with a higher level of complexity, such as top-level subtitling (feature films etc.), and transcreation, remain in high demand and with good tariffs. I believe these types of tasks won't be replaced by AI any time soon.
 

I am a huge fan of Artificial Intelligence and its development. I believe that it is one important reason why my workload as a full-time translator has increased over the last two years. Especially because I chose to extend my services by including post-editing, which is a rapidly evolving area over the last year.
 

Most technology is focused on productivity and efficiency, helping translators do more output in less time. Support systems for job assignment, file transfer, and payments have all helped streamline the indirect work. I am leery of AI translation systems, and have many requests for PEMT work, but so far have not seen my workload change because of it. I generally avoid PEMT work because it pays less but is nearly as difficult and complex as straight translation.


Industry report, 2022

 

This is an excerpt of the most recent ProZ.com industry report. To read the full report, you can go to https://www.proz.com/industry-report/

ProZ.com industry reports are periodic publications that take a look at trends, challenges, and opportunities in the language services industry, from the perspective of the freelance language professional. ProZ.com paying members enjoy immediate, full access to these reports.

 

 

 

 

Everyone is automating what they can.  Automate your marketing by letting ProZ.com do the work of bringing new clients and collaborators to your profile-- just make sure to fill that profile with everything that shows what makes you different from the rest.

If you are not yet a ProZ.com member, you can start now, with special discounts on membership and training until December 15th. Find out more »

 

Topics: translation, interpreting, language services industry

Jared Tabor

Written by Jared Tabor

Jared oversees Member services at ProZ.com. An ex-language teacher, he has lived and worked in Argentina since 1996. He has been with ProZ.com through the La Plata office since 2007. You can follow him on Twitter, @taboredinc .

Subscribe to Email Updates

    Lists by Topic

    see all

    Posts by Topic

    see all

    Listen to the ProZ.com Podcast

     

    Recent Posts