“Jobs and translators that will disappear after 30 years”, “Human vs AI translation confrontation... the gap seems to disappear within 2 years”... These are the titles of articles I've come across often since. This shows that machine translation is evolving at such a rapid pace. In fact, we enjoy a convenient daily life due to advances in machine translation and artificial intelligence (AI). Even if you go on a trip to a language you don't know at all, you can use simple travel conversations without any inconvenience because only a translation app translates the menu board and translates what the locals say in real time. Through these experiences, people think that the “language barrier” has been broken thanks to AI translation.
Of course, it's true that AI translation is growing rapidly. Beyond machine translation, which simply listed matching words in the past, neural machine translation (NMT) now collects sentences that people actually use through neural machine translation (NMT) and translates them into more natural sentences. Researchers and developers in the field of machine translation say that the pace of technological development is so remarkable that machine translation can soon replace human translation.
In contrast, translation experts, more comprehensively, in the translation industry ask, “Is that really possible?” There are many people who make questionable voices. I'm not arguing that machine translation will never replace human translation just to avoid losing jobs.
In fact, “human” translators are responsible not only for the final inspection of the AI translation currently being developed, but also for each stage of supervision. In other words, AI translation development companies are also required to have translation experts on duty. This is because neural network machine translation, which provides more natural translation results, eventually requires humans to confirm whether the results are really natural. This is an essential part of the technology development process. In the end, it's a natural process if you think about what people do and do for people.
Human translators don't just check whether the words in the starting language are well transposed to the corresponding words in the destination language. It also checks contextual meanings that are invisible to the eye but can be felt by humans. It's about understanding the meaning contained in a space between sentences, between the dots in the first sentence, and the meaning of words that vary from sentence to sentence.
In fact, there are quite a few cases where machine translation is an error because it simply determines that something that is used more frequently is correct based on a database. For example, if “LG True Steam” is added as “LG True Steam,” machine translation may be translated as “LG Truth Team.” Since the translation is based on data held by the machine, it was determined that the combination of “Truth” and “Team” is more accurate than the product name “True Steam,” which is a proper noun. But a human translator wouldn't make this kind of judgment.
The role of human translators is still very important in order to improve and advance technology in this way. However, another reason why machine translation is difficult to replace human translation is 'user perception'. In particular, the generation that relied only on human translators for translation and the generation that went through final inspection and confirmation by a human translator (even if using machine translation) would have to go through final human confirmation (regardless of technological development) to have confidence in quality. Therefore, the higher the importance of translation results, the higher the dependence on inspection by human translators. Maybe in that case, people are still doing the translation themselves without considering machine translation itself.
Why are you still researching and developing machine translation? This is because machine translation is not intended to replace human translation; it is a means and a convenient “tool” to help humans. Just as all technologies emerge and develop for human convenience and a better life, machine translation may be an option depending on the situation, but it will be difficult to completely replace the job of a “translator.”
Also, human translators do not think of machine translation as an enemy that takes their jobs, and they should study various ways to use machine translation to obtain better translation results. As machine translation evolves, human translators must also strive to become irreplaceable or better experts than machine translation by developing individual expertise and features.
*This content has been transferred from Gicon Studio to Letterworks.
