Google expands AI language support to reach billions globally
Summary and headline written by AI from the source article. How we work
Google is extending its artificial intelligence technologies to encompass more than 300 languages, reaching over 7 billion people, 86% of the world’s population.
For decades, technology has favored dominant languages, but Google aims to include thousands of languages and dialects in the digital world. The company evolved its translation process from simple text translation to “native audio intelligence,” training models like Gemini to directly process and understand spoken language, capturing nuances like code-switching and emotion.
Google is prioritizing data collection through partnerships with local communities. The WAXAL project, for example, created a speech dataset covering 27 Sub-Saharan African languages. Project Vaani in India is mapping linguistic diversity by collecting speech from over 155,000 speakers across 109 languages. These efforts are supported by tools like Language Explorer, which visualizes the world’s largest open-source language data repository. To reach people with limited internet access, Google developed TranslateGemma, lightweight translation models that function on-device without a connection.
They also support initiatives like Viamo’s “Ask Viamo Anything” which delivers AI-powered information to feature phones, having already answered over 2 million questions in Rwanda. Google is also designing for accessibility, with features like Sign Language-to-Text translation available in Gboard and Live Transcribe, beginning with American Sign Language.
