In the third part of our blog series on Artificial Intelligence, our Content Analyst Lara Armas discusses the use of AI in linguistics.
And what about linguistics?
Nowadays, Facebook uses AI to translate users’ posts into different languages, among other things. Nonetheless, we’ve seen that AI is no match for human translation yet, as it clearly lost the translation competition in 2015 where “accuracy, linguistic expression and logical structure were the key evaluation points”. Today, automatic translations generate drafts of relatively simple language, and for humans it is usually more efficient to correct these drafts rather than translating from scratch. Indeed, although machine learning can be really useful for translation, it still has a few weak spots (techcrunch.com): most tools work word by word, which can lead to serious mistakes. Like when a word has two different meanings, or like when “a word at the end of the sentence determines how a word at the beginning of the sentence should be formed”.
For example: “I arrived at the bank after crossing the street” versus “I arrived at the bank after crossing the river”. “Bank” is not translated the same way in these two sentences, but one has to understand the meaning of the whole phrase to identify the different senses. The same happens with “The animal didn’t cross the street because it was too tired” versus “The animal didn’t cross the street because it was too wide”: here “it” refers either to the animal or the street.
Translation is about meaning, which is what online translators – such as DeepL – slowly started understanding. So ultimately, AI can help professional translators with their work: machine translation will do the main part and humans will proofread the outcome. Robots will likely have a hard time fully understanding cultural references, political and historical contexts. They will struggle to translate words that have no direct equivalent in other languages, or gather the evolution of a word’s meaning. Moreover, it is “not yet capable of the creativity, understanding, and personality that make for truly effective translation, localization, or transcreation.” So far, language is simply still too personal and emotional for computers and machines.
But as some of the tech giants are intensely investing in this sector, we might witness a breakthrough in less than 10 years. Indeed, AI is one of the “hottest spaces in tech right now, with just about every major tech company scrambling to employ it in some fashion, be it to build that perfect digital assistant or to teach a car how to drive itself” (mashable.com). Baidu, IBM, Google and Microsoft all are eager to be leaders in this field and the budget invested will increase from $7 billion last year to over $ 16 billion by 2021 (venturebeat.com). So far, they’ve launched “an earpiece that can translate languages in more-or-less real time” and “a dictation app in 20 languages that could […] provide real-time translation in 60 languages” among other inventions.
As Forbes puts it: “It is as concerning as it is amazing – simultaneously a glimpse of both the awesome and horrifying potential of AI”. Some of the smartest (and richest) minds on the planet disagree about the threat that AI poses to humankind: Mark Zuckerberg, Bill Gates and Demis Hassabis (“leading creator of advanced artificial intelligence” and “co-founder of the mysterious London laboratory DeepMind”) disagree with Elon Musk – CEO of Tesla, SpaceX, PayPal and, among other things, leading doomsayer – about the dangers of AI. I, for my part, am very optimistic about the advances of artificial intelligence and how it is going to improve our lives!
Look out for the forthcoming publication by our founder Hannes Ben, further detailing the role of Artificial Intelligence in localisation strategy.