Meta has just developed an artificial intelligence capable of instantly translating 200 languages. An important step in creating a universal real-time translator that could have multiple uses.
“Imagine visiting a favorite Facebook group, coming across a post in Igbo or Luganda, and being able to understand it in your own language with the click of a button.“For Meta, the group that owns the Facebook, Messenger, Instagram and WhatsApp applications, this reflection was the starting point for the project. No language left behind (NLLB), which aims to create a supertranslator capable of instantly translating all the languages of the world. He has developed an artificial intelligence (AI) which is already capable of translating 200 languages. It’s a giant step, but NLLB-200 is still far from being able to translate the 7,000 languages spoken in the world.
Meta AI: minorities are not left behind
Meta Group’s new AI is based on the machine learning (machine learning in French), a technique that makes algorithms discover recurring patterns in datasets available on the web. It innovates because it does not go through English to carry out the translation. This is because most machine translation systems first convert one language into text, then translate it into the second language before converting the text back into speech. This process delays the translation process and creates a great dependence on the written word, which limits the effectiveness of these systems for mainly oral languages – such as dialects for example. Meta works on a direct speech-to-speech system so that the translation process is faster and more efficient. The task is difficult, because some languages have little data on the Internet, which complicates the learning of AI.
NLLB-200 has already managed to integrate 200 languages, some of which are minority. And 55 of them are from Africa, which allows the translation of languages like Kikuyu – spoken by 6 million people in Kenya – or Kamba – spoken by less than 4 million people – with very good results. Compared to other artificial intelligence translators on dialects from Africa and the Indian subcontinent, Meta’s AI is 70% more efficient. It will therefore be able to reduce the global digital divide between Europe and North America on the one hand, and the rest of the world on the other, which is particularly visible on the various Wikipedias. For example, there are only about 3,260 Wikipedia articles in Lingala, a language that has 45 million speakers in the Democratic Republic of Congo, Central African Republic and South Sudan. In comparison, there are more than 2.5 million articles in Swedish, while this language is spoken by only 10 million speakers in Sweden and Finland.
Connect users around the world to social networks
Meta’s new AI is available as open source for Wikipedia contributors and linguistics researchers and developers. As for the American group, its goal is to make Facebook and the rest of its social networks accessible to more people. This could be a good solution to remedy the group’s loss of momentum, which is facing strong competition, mainly from TikTok, while the firm is developing its Metaverse, which will require real-time translations.
“This is going to be especially important when people start teleporting through virtual worlds and having experiences with people from different backgrounds.“, explained Mark Zuckerberg. “Now we have the opportunity to improve the Internet and set a new normal where we can all communicate with each other, no matter what language we speak or where we come from. And if we are successful, this is just one example of how AI can help bring people together on a global scale.“
The American group revealed a demo called Stories Told Through Translationwhich uses the latest AI advancements from the project No Language Left Behind. This demo translates books from their original languages, such as Indonesian, Somali, and Burmese, into other languages for readers. An excellent way to discover tales and legends from other cultures, and to bring together populations unfairly separated from the rest of the world because of their linguistics.