The Impact of Artificial Intelligence on the Language Translation Services Industry

A couple of days ago my 8-year-old daughter asked me how to say a word in French. Since I try to make her find out the answers by herself, I told her she could look the word up in Google Translator. “Right,” she said, “Google knows everything”. I must confess that her reaction alarmed me a bit. She used the application, but the reply drawn was not appropriate to the context in which she was writing for her school homework. And that alarmed me even more.

Certainly, machine translators—built on complex Artificial Intelligence systems—reduce and simplify basic, common, large tasks, and are able to learn over time. Nevertheless, despite their capacity and high efficiency both in terms of time and money, they don′t possess determined skills exclusive to human beings, such as intuition. For instance, when I need to write a full text in, let′s say, French, I often use a software to start with, and then I correct or adapt it manually. I can do that because I speak French. Professional translators employ more sophisticated tools, yet, they proceed similarly.

To analyze the impact of using AI systems in the language translation services industry, and beyond, I have identified three dimensions:

1) Cognitive Dimension

First, language translations have seen major advances through applications that recognize image or voice —such as Google′s Word Lens— while decoding in real time. Other devices “will even speak the phrase for you,” according to opinionist Conner Forrest in his 2015 editorial “The First 10 Jobs that will be Automated by AI and Robots”, appeared on www.zdnet.com. Thus, if you're not translating complex documents, highly developed algorithms will most likely perform a good job for you. But when it comes to more sensitive fields such as security, medicine or technical, there is substantial evidence that computers are unable to translate 100% accurately, potentially generating a risk with serious consequences.

2) Psychological Dimension

Second, and on a broader scale, some jobs also require mastering certain skills or conveying that “comforting feeling that someone is paying attention to your needs” as serial entrepreneur, technical innovator and futurist Jerry Kaplan (2017) explains in his publication entitled “Artificial Intelligence: Think Again”. Machines are not programmed to have feelings or show emotions or empathy. This adds a psychological component to the cognitive limitations that AI has. Let′s put it in context: as a potential client for a translation company, you would agree with me that you prefer to describe your requests to a person, rather than to a robot. After all, we are human beings with human expectations.

3) Cultural Dimension

Third, there are still cultural signals that technology misses. Take Spanish (or more specifically Castilian) as an example, which is the majority language spoken in 21 countries, totaling around 470 million native speakers worldwide*. One word in Castilian may mean one thing in Argentina and have a completely different denotation in Spain, Mexico, or its neighbor Chile. Even if the software offers different variations and can adapt to local nuances, it is very likely that it will get it wrong.

Artificial intelligence should be understood and used within the limits of cultural and social contexts. Kaplan believes that “Social and cultural conventions are an often-neglected aspect of intelligent-machine development”. He also highlights the importance of what he calls emotional labour and other job-related skills that AI systems are not able to perform or develop. Computers, he argues, don′t have minds and there is no relevant evidence that that will change one day.

We fear that intelligent devices will disrupt the labour market causing mass layoffs. A study conducted in 2013 by the Oxford Martin School in the United Kingdom estimates that approximately 47 percent of all the jobs in the United States are jeopardized by some level of automation (Forrest, 2015). There are relevant available data confirming this global trend.

However, some experts claim that robots will rather reallocate jobs than displace them (see “Robots Will Take Your Job; Will They Guarantee Your Income?” published in 2016 by Scott Santens). Santens divides work in 4 types: routine, non-routine, cognitive and manual. For him, humanity is heading to a world where “jobs are for machines and life is for people,” strongly affecting national economies. In the long term, this impact will be an economic gain for society.

Embracing AI

For all these reasons, cognitive computing should be embraced by humans, not feared. We should do our best to maximize its power even when we think the outcomes could be dangerous, even when the computers may become smarter than us, because in the end, they will do what we want them to do, as Dominic Basulto (2015) reckons in an essay published by The Washington Post entitled “Don′t Fear Artificial Intelligence”. The specialist criticizes those that center the debate exclusively on people versus machines, while he prefers a scenario where both merge into a collaborative form of life. For him, human beings will keep the control with enhanced competencies thanks to technology. As a result, several new sorts of human jobs might emerge, entirely fresh roles that complement the tasks performed by artificial mechanisms.

Furthermore, there are currently concrete research directions that can be pursued to help maximize those benefits that AI brings to mankind. In “An Open Letter on Artificial Intelligence,” signed in 2015, numerous scientists, economists and experts in the field declare that this technology has the potential to eradicate disease and poverty, among other social justice causes. Still, they warned that research should not focus only on the benefits but also the potential negative effects in order to eliminate or reduce them.

So, would Google Translator –or the like—be able to substitute a human translator? In my opinion, the answer is a resounding no. As much as we are aware of the tremendous benefits that translation applications (and by extension AI) bring to the translation services industry and our lives in general, these systems will not be able to fully replace the job of a human translator. He or she will still be in charge of deciding which word is the right one, depending on the context.

My conclusion is that machines can′t express emotions, act intuitively or grasp cultural cues, they can only automate tasks, not jobs. And we should all be prepared to take advantage of that.

*Academic Department of the Cervantes Institute. (2014). El Espa?ol: Una Lengua Viva. Instituto Cervantes.