Google has just expanded its translation capabilities significantly, adding 110 new languages with the help of its PaLM 2 AI model. This update brings the translation tool to over 614 million people, approximately 8% of the world's population.

This expansion covers a wide range, from widely used Cantonese to Manx, a language that was once extinct on the Isle of Man. Interestingly, Google chose the older PaLM 2 model to accomplish this task instead of the newer Gemini AI series. We have reached out to Google for clarification, and if they respond, we will update this article.

African languages have seen a significant boost, accounting for a quarter of the newly added languages. Fon, Kikongo, Luo, and Wolof can now be used, greatly increasing Google's coverage of African languages.

Some interesting additions include:

Cantonese: It has long been requested, but adding it has been challenging due to its similarity in written form to Mandarin.

NKo: A standardized West African language that uses a unique alphabet invented in 1949.

Tamazight: A widely spoken Berber language in North Africa, supporting both Latin and Tifinagh scripts.

Google focuses on the most common forms of each language while acknowledging the wide variations that exist. For example, their support for Romanian is based on Southern Vlach, but it also incorporates elements from other dialects.

This is Google's largest language update to date. In a blog post, Isaac Caswell, a senior engineer at Google Translate, explained, "We're using artificial intelligence to expand the range of languages we support." PaLM 2 is crucial for efficiently learning related languages, such as Hindi variants and French Creole.

This update aligns with Google's goal announced in 2022 to support 1,000 languages. At that time, they added 24 languages using zero-shot learning, which means AI translates without examples.

While impressive, challenges still remain. Machine translation continues to struggle with dialects, cultural contexts, and evolving language usage. Google plans to continue collaborating with linguists and native speakers to improve its system.