Fundamentally, translation is a harder problem than it gets credit for. At the dictionary level, you're mostly alright, as words and concepts tend to have prett

Fundamentally, translation is a harder problem than it gets credit for. At the dictionary level, you're mostly alright, as words and concepts tend to have pretty close direct translations (though you'll often need context, as homynyms exist). A step above that, forming full statements, you run into some difficulty, due to the difficulty in translating the subtleties of word choice, as you've got a fair amount of cultural context to take into account. "Forgive me Father for I have sinned" does not carry the connotations as "Sorry, Daddy, I've been naughty". But a step up from there, translating entire texts, you need a pretty complex theory of the mind, as you need to take into account the context of the author's perspective, their interpretation of their intended audience's perspective, and your audience's perspective. For example, my "Forgive/Sorry" joke earlier relies on the context of both you and I being somewhat aware of A) the Catholic church's process of confessional and B) the relatively modern use of "Daddy" and "naughty" with sexual connotations, which, forget language specific, is culturally specific, where the joke would break if you tried telling it to, say, a place that still used "Daddy" as perfectly normal way to address one's father, or lacked the cultural norm of sexualizing authority. If someone was trying to translate this comment to another language, they might have to completely alter that joke in order for it to make any sense, at which point that entire last paragraph would have to change, etc. Modern AI tools have largely reached a point where the homonym problem isn't crippling them anymore, but haven't really reached much beyond that.