By accounts (I have never experienced), Waymo works a lot better than I expected. But I was still not wrong. Waymo has a remote operators who intervene. As long as that intervention is required, self-driving cars don't really exist.
You can have a modest taxi fleet, but you can't scale that too much. Personal cars definitely won't function like that.
And even if Waymo works, it is still a research project. It isn't a viable business model. That doesn't say anything about whether the technology works, precisely, but the dream of replacing low paid drivers with robot cars was always a bit dubious. Drivers are cheap!
True self-driving vehicles would be a neat invention, but the case for 96% self-driving is less obvious.