Automatic translation for neglected languages

© iStock.

Meta (previously known as Facebook) developed an AI model capable of translating across 200 languages, many of which are not supported by current automatic translation tools that usually only cater to a handful of languages. Many African and Indian languages that have little or no commercial appeal are under consideration. So that other researchers can build on the work, Meta opened the model’s access code in Python and especially the dataset used to train its model. What’s more, Meta offers grants to researchers who wish to contribute to its project to automatically translate “low-resource” languages but who struggle to find funding. “I think that’s really, really important, because it’s not like one company will be able to holistically solve the problem of machine translation,” said Angela Fan, a Meta AI researcher. “It’s everyone — globally — and so we’re really interested in supporting these types of community efforts.” The open code for the “No Language Left Behind” project is available on the GitHub platform.

⇨ The Verge, James Vincent, “Meta open sources early-stage AI translation tool that works across 200 languages.”

2022-07-06