Languages have always been one of the biggest barriers for mankind, because of the difficulty of communication. That being said, languages also are a very important part of this world, because it brings centuries of culture along with them. Languages are not only a way of communicating, but each language has special words with different meanings. Because of how complicated a language can be to interpret, the role of a translator is very important. The first ever translation dates back to when Buddhist monks were translating Indian texts into Chinese. From then on, translation has evolved leaps and bounds. Today I am here to talk about the newest and most advanced form of translation, which uses AI, and is called Neural Machine Translation (NMT).
As a student at a school where I have classes in both French, English, and Spanish, translation apps have been a big part of my life for homework, writing emails, etc. Although the language I speak at home is English, sometimes when I am writing an email in English, I forget a term, and need to use a translator to translate from French to English. Ever since I was a young child, Google translate had been my go to translator because it was the most popular, and because of the camera function, which was very useful for translating entire paragraphs of paper. But lately, I have been using a new translator, called DeepL, which uses NMT.
Google translate was first released in 2006, and it used statistical machine translation to function. Although this form of translation was pretty convenient, this translator lacked accuracy, and the translations weren’t exactly perfect. Either way, this tool was and still is extremely useful, with over 500 million users by 2016 alone! In 2021, there had been 1 billion installs of the app!
With DeepL, AI in the translation industry really took off in 2017. The data used to train this neural translation system came from Linguee, a multilingual lexicon that compares translations in close to 20 languages.
And so the birth of NMT began. This artificial neuron-based, intelligent technology takes the full text and its context into account. The waterfall of data NMT receives also enables it to continuously enhance and perfect itself. Ever since then, many other apps have integrated NMT translation, such as Google Translate, Microsoft Translator and Meta’s translator.
So what are the Pros and Cons?
Well, there are many advantages. AI makes it possible to translate large volumes of text very quickly while maintaining accuracy and precision. Machine learning-based translation software with AI capabilities has the capacity to self-correct and raise the calibre of the translations produced. Many AI-enabled translation software can also translate several texts into various languages at once. Finally, these techniques provide consumers the benefit of being extremely affordable (or even free) and covering a very broad range of languages.
Despite all of its benefits, AI still has a lot of shortcomings when it comes to translation. In fact, this technology is unable to modify the translation for the intended audience. Additionally, it cannot take into account regional cultural norms and practices, customer expectations, the translation’s style, or its intended use. These are crucial components in translation in order to produce writings that are true to the original, faithful to the intended audience, and respectful of the local culture. In addition, while AI translation is capable of being efficient for the most widely spoken languages (such as English, French, Spanish, German, Dutch, Italian, Arabic, etc.), it is far less efficient for dialects or rare languages for which there is little available data. The AI will frequently be forced to apply English translation in such circumstances.
But in my opinion, AI translators are great tools, and are much more precise and accurate then statistical machine translators. I’m sure that in the future, AI will be able to solve the disadvantages that it possesses in the moment.
Thanks for reading this article, hope it was enjoyable, and tune in next time for more!