Most often when we think about the difficulties associated with the translation of a language, we are thinking in one direction: you are learning a new language, and it is difficult to translate your own language and thoughts into this new one. This difficulty feels extremely real and visceral, and engages generally with many aspects of your overall cognitive experience: it’s difficult to remember words, grammar rules, it’s embarrassing to make mistakes, frustrating to not feel understood, etc.
But there is also another direction, which we talk less often, and that is a bit more subtle: it’s also difficult to speak your own native language, to someone who doesn’t speak it well. When you do that, you need to monitor and adapt your own language level, and be on the lookout for clues that your interlocutor might not fully understand what you say. You need to be patient, empathic and observant. That is why, in an interlingual couple that has settled, for a long time, on an intermediate language (English in my own case, as the link between Persian and French), it can be surprisingly hard, in the context of daily life, to “switch to another language” momentarily, even for fun, or learning purposes.
With AI, as we all know, this difficulty essentially vanishes completely, because it is extremely good at picking up, without any trace of doubts or hesitation, any verbal stuff that you might throw at it (even in foreign languages!). AI is, in a sense, the ideal interlocutor, but it also comes at a price, because the way it currently works, you cannot be completely sure that it is not overly acceptant of you. Perhaps in the future there will be some kind of “knob” that can you adjust, for that? Or better yet: you will be just able to ask them (to be less sycophantic), and you will have good reasons to believe them (that they will). But could it really work, or is there a deep reason why it couldn’t?