There is no doubt that the development of machine translation, both written and oral, has reached an impressive level and it is safe to say that this trend shall continue in the future. Despite the growing number of applications and devices enabling us to communicate with people who do not speak our language, it seems that this type of communication usually boils down to the most basic and simplest forms of communication. Even though devices can be connected to one another quickly and easily, connecting people from different parts of the world is still not as easy, due to  language barriers persisting despite all available assets.

The difficulties in translation performed by artificial intelligence are most obvious in machine translation services, i.e. GoogleTranslate and Facebook’s automatic translation service, which is used in 4.5 billion translations each day. These difficulties are most often reflected in harmless mistakes that are usually easy to recognize and represent nothing more than a short-lived source of humor and fun. However, such errors are occasionally found in a broader context where they can have serious legal, economic, and other consequences. The Israeli newspapers Haaretz published a story about a Palestinian construction worker who was arrested near Jerusalem after posting a picture of himself standing next to a bulldozer on Facebook accompanied by a status in Arabian that read “يصبحهم”. Facebook’s automatic translation service translated his status as “Attack them” in Hebrew and “Hurt them” in English. The police was alarmed and the worker was soon arrested on suspicion of plotting to commit a terrorist attack using the vehicle. He was released a couple of hours later, once the police noticed the error: the original status in Arabian simply said “Good morning”. Errors of this kind point to deep-seated issues with machine translation services. It is known that random substitutions of toponyms and personal names (London becomes Amsterdam, Samsung becomes Motorola) are a frequent occurrence in machine translation. However, if a quite common phrase (“Good morning”) is translated incorrectly not in one, but in two languages and due to the particular sequence of circumstances compromises the safety of several people, to what extent can we rely on these services for the translation of more complex phrases and sentences?

The ideal scenario for machine learning and artificial intelligence would be one with a clear set of rules and precise indicators of success and failure. Chess is a great example, which is illustrated by the fact that artificial intelligence managed to beat the best human chess players much faster than expected. On the other hand, language and its unlimited possibilities are a much more difficult task. For this reason, global markets are still waiting on reliable machine translation. Despite recent advancements and the fact that machine translation is capable of processing whole sentences as units and not only translate individual words, translation errors are still common and we are still waiting for a time when automatic translation will be capable of consistently providing an accurate translation of the source text. The problem is that it is not enough to look at sentence as a whole. Just as the meaning of a word depends on the rest of the sentence, the meaning of the sentence depends on the remainder of the chapter and the text while the meaning of the text depends on the broader context, the speaker’s/writer’s intentions, etc. Idioms, irony, and sarcasm, for example, only make sense in a broader context, and understanding a text involves a variety of intertwined activities that we perform without thinking and that, for now, cannot be replicated by a machine.

Nevertheless, machine translation available today has undisputedly proven to be an extremely useful tool being applied in various contexts. For example, the translation of web browsers or orders when shopping online. The most important thing to keep in mind is that for the time being machine translation should only be used as a tool in specific environments and not as a substitute for translators. That day is still in the distant future and it is uncertain if it will ever come. We don’t know what the future holds, but many experts agree that further development of machine translation will greatly transform and facilitate the work of human translators without actually replacing them.


Braće Popović 5
21000 Novi Sad, Serbia

Contact Phones

Phone: +381 21 47 25 227

Fax: +381 21 47 25 226