The company introduced an artificial intelligence model that supports over 400 languages as the first step of this project. This model was described by Google as "the widest language coverage ever seen in a speech model." Language and artificial intelligence have been at the core of the American tech giant’s work for some time now. Recent advancements in machine learning have also led to significant developments in multifunctional large language models and the emergence of new products.

Language models can be used for many different purposes such as in search engines, generating content with linguistic structure, and translation. Google's new 1000-language initiative aims to be used not for a single function, but as a single system with extensive knowledge about the world's languages for various purposes.

Zoubin Ghahramani from Google's artificial intelligence division, who spoke to The Verge, stated that the company believes such a large model could provide various AI functions for languages that are not widely available online. Ghahramani expressed the following on this matter:

“A model that is exposed to and trained on many different languages enables us to achieve better performance in the underrepresented (low-resource) languages we are talking about. The path to reaching 1000 languages is not by creating 1000 different models. Languages are like organisms; they have evolved from one another and therefore share similarities. When we include data from a new language into our model, we can obtain translations from common languages to low-resource ones and make significant progress.”

The company also adds that it will finance difficult-to-access data, such as written texts and voice recordings, required for low-resource languages. It’s worth mentioning that Google does not have a direct plan for how it will apply the functionality of this model. For now, the tech giant expects it to be used in various Google products, such as Google Translate and YouTube subtitles. Ghahramani also notes that these types of large language models have huge potential because they can perform a wide range of tasks, such as translating between languages, converting code into commands for robots, and solving math problems.

A similar project also came from another tech giant, Meta. Led by Mark Zuckerberg, the company announced earlier this year that it is working on an AI project, referred to as the "universal speech translator," which aims to eliminate language barriers both online and physically.