I agree with yomikoma about using LLMs for translating documents… the potential for inaccurate information seems really high and there's stuff that a fluent (or even not-native level heritage speaker) human can intuit but a computer will just steamroll right through.
no subject
I agree with