After the hefty €1.2 billion fine for breaking the GDPR, Meta aims to win back public opinion with an ambitious project based on iartificial intelligence (AI).
It is a project that aims to preserve the linguistic diversity of the world thanks to the Massively Multilingual Speech AI models, capable of recognizing over 4,000 spoken languages and transforming them into written text and vice versa.
Meta’s new AI project starts from the Bible
In a post on its official blog, Meta underlined the risk that many minority languages, belonging to small communities and little known by the rest of the world’s population, could disappear over time.
Use ChatGPT with NordVPN – 56% off annual plan + 3 months free
To give them voice and access, Meta relies on its templates Massively Multilingual Speech (or MMS). Through AI and machine learning, Mark Zuckerberg’s company has collected audio data from thousands of different languages, mainly based on religious texts translated into various languages, such as the Bible for example.
By comparing the translations and the original texts, Meta created a database of New Testament readings in more than 1,100 different languagesthen reaching 4,000 with other Christian documents.
Although the content of the audio recordings is religious, it is clear that the model can also adapt to other types of content. These are still solutions to be refined and refined better. However, they represent a solid basis for protecting “threatened” languages.