Facebook's AI Engine Can Now Translate 100 Languages

Ilustrasi Machine Learning Facebook (dok. Facebook)

JAKARTA - Facebook is improving its machine learning capabilities, so that it can translate 100 languages. This allows Facebook to be well received by 2 billion users worldwide.

"This is a milestone, the pinnacle of Facebook's AI work for machine translation for many years," Facebook research assistant Angela Fan said in a blog post, reported by Engadget, Wednesday, October 21.

Angela claims the use of machine learning developed by Facebook can translate language more accurately than previous tools. Where before, Facebook's machine translator relied too much on English as an intermediate language before being translated.

Now users can translate, for example, Chinese to French, without the need to use intermediate Chinese to English, then English to French, to get the results.

Facebook says it handles an average of 20 billion translations every day, and hopes the new system will deliver better results.

"Breaking the language barrier through machine translation is one of the most important ways to bring people together, provide official information about COVID-19, and keep them safe from harmful content," said Angela.

Artificial intelligence technology is also an important point in machine learning implemented by Facebook. So that it can translate words in a new language.

For example, the language combination between Bengali, Hindi, Marathi, Nepal, Tamil, and Urdu is quite difficult to translate. This includes adding new vocabularies that are rarely translated, such as Icelandic-Nepalese or Sinhala-Javanese.