Yeah, I agree - I get what the author is saying, but I also don't expect "translator" to be a practical career path in the future.

Even small, dumb, local models are excellent at translation already. Frontier models are on par or better than the human translations we've tested them against at work.