"If they'll keep consuming AI-driven media, they'll take that as normal and won't blink twice if that becomes the de-facto standard in about 10 years."

There is a sense in which that is true.

However, we all develop taste, and in a hypothetical world where current AI ends up being the limit for another 10 or 20 years, eventually a lot of people would figure out that there's not as much "there" there as they supposed.

The wild card is that we probably don't live in that world, and it's difficult to guess how good AI is going to get.

Even now, the voice of AI that people are complaining about is just the current default voice, which will probably eventually be looked on about as favorably as bell-bottomed jeans or beehive hairdos. It is driven less by the technology itself than a complicated set of desires around not wanting to give the media nifty soundbites about how mean (or politically incorrect) AI can be, and not wanting to be sued. It's minimal prompt engineering even now to change it a lot. "Make a snappy TikTok video about whatever" is not something the tech is going to struggle with. In fact given the general poverty of the state space I would guess it'll outcompete humans pretty quickly.