LLMs vs Google Translate

Reminds me of this article by Douglas Hofstadter: https://www.theatlantic.com/technology/archive/2018/01/the-s...

Did you just step out of a time machine from 10 years ago? Translation is considered a fully solved problem due to LLMs. If you're bilingual you can find out for yourself, LLMs are able to parse meaning and transfer writing styles perfectly well. You're able to prompt it with the context of what its translating and it understands that it has to return short meaningful text. I know that it might make mistakes, thats obvious, but considering the post we're replying to, an open sourc

Well, these days, Google Translate is pretty bad as machine translation services go... it's much better to ask an instruction tuned multilingual LLM to translate for you.

People think LLMs are better than Google at translation but it's not true.

Google Translate doesn't hold a candle to LLMs at translating between even common languages.

Google Translate turning a non-word in a source language into an unrelated word in the target language is an old, known misbehavior. I don't think LLMs changed anything there.

LLMs are really good with translations.Google Translate is relatively awful. I have an intern now who barely speaks my native language but very bad English so weve been using it all the time, and its always spot on, even for phrases that dont translate directlyI bet I can do a good job communicating with you without speaking a common language.

Language models are actually really good at translation. In many cases GPT-4 is much better than google translate.

Google translate has huge amounts of training data. I'm not really sure what you are suggesting.

Google translate doesn't understand what it translates either. It just relies on a very large corpus of parallel texts.