Our CIO at work posted a warning about using ChatGPT on sensitive data. The shocking part was that in the set of examples for why we might be using ChatGPT already he mentioned "for performing a quick fact check", which is insane to me. Who would use the system that is know to just generates likely answers even if they are untrue, for a fact check of all things?!
Machine learning is only as good as the dataset it has, and given that english has a HUGE data set on the internet, its okay at it, but it makes sense that for other languages, its likely not ideal.
An example would is art. Look up one using a smaller data set (e.g fully legal ones where all training data had artist permission) vs ones trained on the larger dataset where legality wasnt a concern. Night and day difference
ChatGPT is actually able to translate the information it learns in one language into other languages, so if it's having trouble speaking Bengali and such it must simply not know the language very well. I recall a study being done where an LLM was trained up on some new information using English training data and then was asked about it in French, and it was able to talk about what it had learned in French.
So much new tech only works in English (for example: image generation like StableDiffusion) and using ChatGPT in your native language and it works great, almost as good as in English (and actually even better for lots of people because they can express themselves better in their native language than English) felt unreal