Will Artificial Intelligence Make Interpreters Obsolete?

I read the news about Waverly Labs’ new in-ear realtime translation device (video below) with great interest. With rapid advances in artificial intelligence (AI), interpreters are sometimes asked if technology will soon replace us. The short answer is that technology definitely assists good interpreters and translators, and may eliminate the need for expenditure of human brain effort on certain large-volume, low-priority projects. But, while I do embrace new technology to assist me in my translation and interpreting work, I’m not the least bit concerned that it will send me to the unemployment line.

Differentiating Between Mainstream and Professional Use of Translation Technology

For years, Google has pioneered mainstream use of translation technology through its Google Translate website and Word Lens applications. Skype has also rolled out a real-time voice translation feature for its audio and video calls. I don’t pretend to understand the technology, but I know they build a corpus of usage and use intelligent analytics to suggest what the AI considers to be the most likely translation for a given word or phrase.

I’ve seen these technologies used with great success for non-critical communications. For example, an English-speaking friend who visited a Spanish-speaking church pointed his smart phone at the projector screen to get an instant translation of the worship song lyrics. The possibilities for chatting with a friend in a foreign language via Skype video are also intriguing, though I have never tried it and haven’t heard how the user experience rates.

Professional translators and interpreters (translators work on documents; interpreters work with live speech) have also put translation technology to good use. Patent translators use the International Patent Office’s online translation tool to access data for highly technical patent translations. Professional translation software can access many mainstream and paid corpora that compares existing translations (for example, a European Union document may be translated into ten different languages and published online) to offer suggested translations for a word or phrase. In the recent past, this type of research would have required the translator to spend hours looking at the various websites and references.

So Why Can’t I Just Take My New Translation Device to the Hospital or Court?

Even with access to a mind-boggling amount of data, AI can only get so far in understanding the nuances of human communication. That’s why I’m all for the use of AI translation for non-critical communications.

 

Let’s take Google Translate to court and see how it fares (my professional opinions along the way):


 

Judge: Good afternoon, Mr. Ramirez, please raise your right hand. Do you swear to tell the truth, the whole truth, and nothing but the truth?

Google TranslateBuenas tardes, señor Ramírez, por favor, levanten la mano derecha. ¿Jura decir la verdad, toda la verdad y nada más que la verdad?

Holly: A bit of a grammar issue but generally fine.

Witness:  Buenas tardes, Señoría. Lo juro.

Google Translate: Good afternoon, Honor. I swear.

Holly: Close enough.

Prosecutor: Mr. Ramirez, where did you leave the Colt 45 that was used in the unfortunate demise of the victim?
Google Translate: El Sr. Ramírez, donde dejaste el Colt 45 que se utilizó en la desafortunada muerte de la víctima?
Holly: No big issues.
Witness:  Se me borró dónde quedó ese maldito cohete, pero si quieren saber, el pinche lana quedó en la cajuela del mueble.

Google Translate: I erased me where you got the damn rocket, but if you want to know, click the wool was in the trunk of furniture.

Holly: Oops. A real interpreter would have said something like: “I can’t remember where that damn piece ended up, but if you want to know, the damn dough ended up in the car trunk.”

Prosecutor: I’m sorry Mr. Ramirez, I didn’t get any rockets and I don’t care to click on any wool.


As you can see, communication broke down quickly once the witness used a little slang. The witness gave a perfectly coherent answer (and this type of answer is entirely common in court testimony), but Google’s rendering garbled it beyond recognition.

One problem is that Google cannot consistently give a dependable translation; a potentially greater problem is that it takes an interpreter to parse out which renditions are correct. It should be clear that paying an interpreter to sit in court all day and referee AI translations–constantly striking incorrect translations and providing alternatives–is a much worse solution than hiring an interpreter to provide a professional interpretation throughout the proceeding. The same can be said of critical communications such as hospital and doctor visits–this is why, even with all the technology available, hospitals still use onsite or remote human interpreters.

In sum, translation technology offers great benefits for the casual user and to aid human interpreters and translators.  But, when a person’s liberty or health outcomes are on the line, even sophisticated AI translation still doesn’t hold a candle to the depth of experience and bicultural understanding that a human interpreter applies to understand and translate the speakers’ communications.

 


 

Content Copyright © 2016, Preciso Language Services

PERMISSION TO REPRINT: You may use any items from this article in your print, blog, magazine or electronic newsletter. But in order to do so, you must include the following paragraph, including a link to www.PrecisoLanguage.com.

“Information courtesy of Preciso Language Services and www.PrecisoLanguage.com, a translation and interpreting firm owned by Certified Translator and Master Licensed Court Interpreter Holly Behl.”

 

One thought on “Will Artificial Intelligence Make Interpreters Obsolete?

  1. Harry Morales

    Agree, for now.
    Its just a matter of time that AI take over interpreting and translator jobs, at least a big portion of it

    Reply

Leave a Reply

Your email address will not be published. Required fields are marked *