Insights August 5, 2020

AI in the translation industry: what is it good for?

Living Word
By Living Word

Artificial Intelligence (AI) is one of the most talked-about topics in the translation industry right now. But what exactly is AI good for when it comes to translation, and where does this leave human translators?

Translation has always been crucial to communication between different regions, and only humans have been able to properly translate texts – until recently. Today, a new kid on the block is challenging the human in her role of translator: namely the omnipresent Artificial Intelligence (AI).

AI translators are digital tools that use complex algorithms to translate written or spoken words. It’s a clever, cost-effective way to translate in the 21st century, and many big companies are already using it for their benefit. The virtual assistant Siri, for example, uses AI-based language recognition software, and many language learning apps, like Duolingo, do the same. Another well-known example is Facebook, whose integrated translation programme is solely based on AI and is considered one of the world’s best AI translation services, producing over five billion translations every day.

This, of course, begs the question: should more companies use AI for translation? We take a look at the advantages and disadvantages of AI translation tools.

The good

We hate to admit it, but AI is a far faster translator than we are. A robot can translate a sentence to several different languages in virtually no time, but a human takes a little longer. Moreover, AI never gets tired, but can work endlessly without having to take a break. Another benefit: it’s super easy to use. With a simple click, you can get a good, instant translation.

The bad

A machine is a machine – and machines make mistakes. A recent example: due to a technical error, the Chinese President’s name, Xi Jinping, became a swear word when Facebook posts were translated from Burmese to English. Apparently, the name wasn’t included in Facebook’s Burmese language model, and its AI system tried to replace words with similar syllables – which obviously didn’t quite work out.

This goes to show that AI is still struggling to compete with human translators, much due to its lack of real-world common sense. Humans can think for themselves, decide what’s morally right or wrong, and understand subtext as well as regional differences in communication. AI simply can’t, and so it fails to understand human concepts such as sarcasm. This shortfall might be fixed in the future – complex algorithms are already able to translate some advanced sentiments of messages – but it doesn’t look as though this is going to happen anytime soon.

The verdict

While the likes of Google Translate were often ridiculed in the past due to the many errors they made, new translation methods, algorithms and products are constantly being developed. The gap between human and machine translations has narrowed significantly, and big mistakes are less likely to happen.

Less likely doesn’t mean never, however – and that’s why human translators and transcreators are still absolutely vital. If you’re interested in the quantity of work, rather than quality, AI might be the right option for you. But while it is a great tool that is likely to have a bright future, we don’t advise you to rely solely on AI just yet.

Language has so many nuances and undertones, which AI simply isn’t able to pick up on yet. Humans do. And they give a personal touch to a translation that robots can’t – at least for now.