Google announced a new and ambitious project to develop a only artificial intelligence (AI) language model to support the world’s “1000 most widely spoken languages”which has been identified as the new Tower of Babel, the biblical building and the Genesis account that explains why peoples speak in different languages.
As a first step toward this goal, the Mountain View company is introducing an AI model trained in more than 400 languages, which it describes as “the largest language coverage seen in a voice model today”.
According to a report published on the website of TheVergeGoogle has already started integrating these linguistic models into products like Google Search.
These models are capable of many tasks, from language generation to translation.
Google’s mission: a unique and varied system
Google’s “1000 Languages Initiative” does not focus on any particular functionality, but on create a single system with a wide variety of knowledge in all the world’s languages.
Zoubin Ghahramani, vice president of research at Google AI, told Google AI that the company believes creating a model of this size will make it easier to bring various AI capabilities to languages that are poorly represented in online spaces and AI training data sets. .
“By having a single model that is exposed and trained in many different languages, we get much better performance in our low-resource languages”, explains Ghahramani.
He adds: “The way we get to 1,000 languages is not by building 1,000 different models. Languages are like organisms, they have evolved from each other and have certain similarities”.
The company reveals that it has no direct plans on where to apply the functionality of this model, only that it expects it to have a variety of uses in Google products, from Translate to YouTube subtitles and more.