These things are all on the radar of "innovation" types. I don't mean to say they're not interesting, but in the area of applied ML all this stuff is basically as mainstream as it comes (despite being unsupported by any actual research advances).