Machine translation has made considerable progress thanks to artificial intelligence. Google Translate now supports 243 languages, and Meta has launched the 'No Language Left Behind' (NLLB) project, which covers over 200 languages, many of which had never been processed by an automated translation system.
UNESCO estimates that about 40% of the 7,000 languages spoken worldwide are in danger of disappearing. AI can play a role in documenting and preserving these languages by creating transcription, translation, and learning tools from limited data – a major challenge for natural language processing models that usually require vast text corpora.
Google DeepMind has used AI to help decipher ancient texts, including cuneiform inscriptions and Linear B tablets. This work, published in scientific journals, shows that deep learning models can identify patterns in ancient scripts and propose plausible decipherments.
Real-time translation tools are advancing. Google Pixel Buds and other connected earbuds offer simultaneous translation in several dozen languages, with results that are still imperfect but constantly improving. Real-time voice translation remains a challenge, especially for tonal languages and regional dialects.
Indigenous communities and linguists are collaborating with tech companies to develop tailored translation models. The Masakhane project, led by African researchers, is developing NLP (natural language processing) tools for African languages, a field historically underrepresented in linguistic AI.
