Optimistically, it is both. While biased moderation has turned more than a few places sour, it's certainly not the greatest problem with these platforms. That said, having an algo whose biases are agnostic to individual users isn't a bad thing imho. In theory, it could free moderators/community leaders up to do more constructive things.
On the other hand, this whole thing could be a cash grab for AI training data money.
On the other hand, this whole thing could be a cash grab for AI training data money.