Explanation
The format survives every trend cycle that is supposed to replace it because a human face in frame is the strongest attention signal short-form has, and because it is the only format where a complicated argument can actually be made. A montage can show you something. It cannot tell you why you were wrong about it.
The corollary is that a talking head video is a script with a person attached, and weak writing is not rescued by good delivery. When one underperforms, the fix is almost always in the first sentence rather than in the lighting.
The common mistake
Assuming it needs a real person now that avatars exist. What the format needs is a consistent face, which is a different requirement, and an account whose presenter changes every post never accumulates the familiarity the format runs on.