English on non-English NLP and machine learning projects

less than 1 minute read

Whenever I ask a bilingual (English + another language) students or professionals working on machine learning or artificial intelligence if they considered doing a project in AI for their nonEnglish mother tongues language, such as Hindi or Spanish or Arabic, they look at me puzzled and surprised. Yes, there are a lot of publications on all sorts of languages but how often do you see innovative products in the market for non-English customers even in English-speaking nations? The US has a huge immigration population and houses neighborhoods that don’t even speak English. Why not develop more intelligent products with the aid of deep learning that targets nonEnglish recipients and not just comes up with another translation software every time? We need to think beyond the status-quo of research, products, software, and publications that are predominately English. It is challenging; I admit because it is so easy to code in English with an English programming language syntax, editor, OS, GUI and it is also hard to find nonEnglish corpus. Mandarin is an exception in all this here. It is not impossible to do more for nonEnglish speaking societies.