It's very hard to tell, and ML's UX work certainly could be as intellectually bankrupt as you suggest but my read is they are doing a lot of detailed exploratio

It's very hard to tell, and ML's UX work certainly could be as intellectually bankrupt as you suggest but my read is they are doing a lot of detailed explorations around specific use cases in hopes of informing a larger/broader interaction design suite that can scale across a very wide range of interaction needs, essentially trying to reanalyze the problem space in much the way that led to the distillation of the Windows, Icons, Mouse, Pointer schema at Xerox Parc. Microsoft seems to be keeping its sights much smaller and making much more tactical delta off of existing concepts (eg. much of Windows MR interactions really is just floating traditional windows with traditional mouse and keyboard interactions).