无论您是从美国，巴西，婆罗洲还是法国登录，Facebook都可以使用自动机器翻译将其平台上发布的几乎所有书面内容翻译成本地语言。实际上，仅Facebook的“新闻提要”每天就提供约200亿笔翻译。但是，这些系统通常使用英语作为中介步骤-也就是说，从中文到法语的翻译实际上是从中文到英语到法语的翻译。这样做是因为往返于英语的翻译数据集非常庞大且广泛可用，但是将英语置于中间会降低整体翻译的准确性，同时使整个过程变得比所需的更为复杂。这就是为什么Facebook AI开发了一种新的MT模型，该模型可以在不使用英语的情况下直接在两种语言（中文到法语以及法语到中文）之间进行双向翻译。

Whether you log in from the United States, Brazil, Borneo or France, Facebook can use automatic machine translation to translate almost all written content published on its platform into local languages. In fact, Facebook's "news feed" alone provides about 20 billion translations a day. However, these systems usually use English as the intermediary step - that is, the translation from Chinese to French is actually the translation from Chinese to English to French. This is because the translation data set to and from English is very large and widely available, but putting English in the middle will reduce the accuracy of the whole translation and make the whole process more complicated than needed. That's why Facebook AI has developed a new MT model, which can translate directly between two languages (Chinese to French and French to Chinese) without using English.

Facebook AI研究助理Angela Fan表示：“最大的挑战实际上是，我们如何利用现有的翻译系统，然后真正满足世界各地人们的需求。“因此，您正在翻译成人们真正想要的所有语言和所有方向。例如，世界上有很多地区人们会说多种语言，这些语言都不是英语，但是现有的翻译系统严重依赖纯英语的数据。”她指出，在Facebook平台上每天以160种语言发布的数十亿篇文章中，三分之二是用英语以外的其他语言撰写的。