I also wondered this before I ever saw any. There is a gradient from someone streaming with no cam, then a non-human vtuber, then human vtuber and finally a real human.

The simplest answer is probably that we all have our likes and dislikes, but personally for me I fall inbetween the non-human and vtuber section. If they stream permanently as e.g. a fish I find it too odd, whereas apparently my brain is perfectly happy seeing any kind of expressive humanoid that I can link to the speaking person.

If they have no cam then its just a voice over which feels too impersonal for the streaming space, which I don't like.