Machine translation, or MT for short, has been around in one form or another for at least the last two decades. The technology is steadily improving and one form or another of it is used by individuals for their own personal use, but is it really going to be good enough to threaten the livelihoods of professional human translators? One comment that has been made about Artificial Intelligence (AI) and how it relates to MT is that when computers know when to laugh and why something is amusing, then it’s time to think about making humans redundant!
No Obvious Timeline for MT Replacing Human Translators
The real answer to the question that is the topic of this article is that nobody really knows when or if machine translation will become so good that human translators are no longer needed. It’s not like building a physical structure like a skyscraper or a bridge. Some translators give it another ten years before MT developers have designed software that can replace human translation. Others say that their translation tasks are too nuanced for machines to ever replace them.
There are certainly commercial reasons for using MT rather than pay humans to do the job of translation. A really efficient MT would cut the cost of translation dramatically, even if a human was paid to do the last bit of editing and proofreading. To some extent that’s happening already. However, one of the lessons of the recent Boeing plane disasters is that introducing technology before it has been perfected is not such a good idea.
The two Boeing 737 Max 8 crashes, one in Indonesia and the other in Ethiopia, are now suspected to have been caused by software developed for the planes’ automatic guidance systems that was faulty and more importantly, couldn’t be corrected by the planes’ human pilots when something went wrong. Part of the reason why these defects were introduced too soon could be blamed on the incessant need to drive profits. Boeing was under considerable pressure t come up with a plane that could match the Airbus.
Implications for Introducing MT Too Quickly
It may be a long bow to draw to make a comparison between aircraft control mechanisms and MT, but there are similarities in the potential devastating effect that a wholesale move to non-human translation and the reduction of control by pilots in their planes. There are several reasons to be cautious about a too early switch to MT in the translation industry. Take the field of medical translation, for instance. Medical translators are normally trained in both translation and medical science and its applications.
A good medical translator could review a translation completed by a human translator and immediately spot glaring errors in translation. There are very serious consequences for getting medical translation wrong. Drugs, for example, are distributed worldwide. Their prescription and use may be restricted to medical professionals, but these professionals, nurses, doctors, anaesthetists and surgeons, must follow instructions provided for the use of the drugs by their manufacturers. If the drugs are manufactured in Germany, or Korea, these instructions must be translated into a myriad of languages for use all around the world.
In reality, the drug manufacturer probably restricts the initial translation to a fraction of this number of languages and relies on further translation in the countries where the drugs eventually end up in. A mistranslation can easily result in very serious medical consequences.
No doubt a serious mistake would result in a recall of the drug, but like the grounding of the Max 8s, it could be too late to prevent ill health and deaths.
Less critical in terms of the effects on human lives, but still very important are translations performed when politicians of different countries communicate with each other. It’s not difficult to imagine the effects of using MT too liberally when nuanced expressions are incapable of being translated accurately by a non-human translation tool.
The Main Advantages of using MT
MT’s main attractions are its speed and associated lowering of cost. A typical document that might take a human translator an hour to complete might only take seconds using MT. Any business or organisation that requires large volumes of translation would make considerable savings if it could rely on using MT instead of turning the task over to a professional translation agency.
MT is being used most frequently at the moment by individuals for personal use. It has a real role to play when people need to get the gist of a text which they cannot understand because it is in a different language from their own. When there is no necessity for absolute accuracy, then MT comes into its own.
In the travel field, for example, the casual visitor to another country, whether for pleasure or business will find ubiquitous translation tools, like Google Translate, perfectly acceptable to understand all the signs and notices that surround them when they travel around. However, even in the tourist industry, too much reliance on the use of non-human translation can backfire, even if the mistranslations are merely comical embarrassing or just plain confusing rather than dangerous.
Travel blurbs in Japanese, for example, about the wonders of the Great Barrier reef, that have been translated by MT may seem funny to the Japanese reader but may backfire as amateurish information may imply that the attraction itself is not worth bothering with.