Can someone answer this point I'm confused about: Why can't a phone generate bokeh blur using the same technique that an ordinary camera does, ie. a large aperture / small f-stop? IOW why is ML or depth sensing needed at all?
First, phone cameras all have a fixed aperture. If you set the aperture wide enough to get bokeh in a typical portrait shot, then the depth of field would be too narrow to be practical for general photography.
Second, the depth of field depends not only on the f-stop but also on the focal length (or, to put it another way, it depends on the absolute diameter of the aperture). Given the focal length of a typical phone camera, you'd need a lens with a very wide aperture to get significant background blur for subjects at portrait distance. Wider apertures complicate the complexity and expense of lens manufacture.
It is interesting to speculate on the use of very wide aperture lenses on future phone cameras, though. If such lenses could be made cheaply, one can imagine that the shallow DOF could be overcome via some form of automated focus stacking.
Also, you probably want the depth of focus to be large for general purpose photography. And you don't have a variable aperture in that phone (I guess).