Meta's Latest AI Can Instantly Translate 200 Languages
Meta has open-sourced an AI model said to be capable of translating among 200 languages.
The company announced(Opens in a new window) in February that it was "building a new advanced AI model that can learn from languages with fewer examples to train from, and we will use it to enable expert-quality translations in hundreds of languages, ranging from Asturian to Luganda to Urdu." It called the effort No Language Left Behind, and the result is a model dubbed NLLB-200.
"When comparing the quality of translations to previous AI research, NLLB-200 scored an average of 44% higher," Meta says(Opens in a new window). "For some African and Indian-based languages, NLLB-200’s translations were more than 70% more accurate." (The company offered more information about the models to which it's comparing NLLB-200 in a separate blog post(Opens in a new window).)
Meta says the improved translations enabled by NLLB-200 "will support more than 25 billion translations served every day on Facebook News Feed, Instagram, and our other platforms." The model can "help to spot harmful content and misinformation, protect election integrity, and curb instances of online sexual exploitation and human trafficking(Opens in a new window)," it says.
NLLB-200 could "help advance other technologies, like building assistants that work well in languages such as Javanese and Uzbek, or creating systems to take Bollywood movies and add accurate subtitles in Swahili or Oromo," Meta suggests. It's also partnered with the Wikimedia Foundation so the model can assist with the translation of Wikipedia articles.
Meta also thinks NLLB-200 will be vital to the metaverse. "As the metaverse begins to take shape," it says, "the ability to build technologies that work
Read more on pcmag.com
