That functionality (first necessary, then required by guidelines, then expected, and then still usual) disciplined UI designers to make things doable in a clear sequence of actions.

Now they think any ape can make a UI if it knows the new shiny buzzwords like "material design" or "air" or whatever. And they do! Except humans can't use those UIs.

BTW, about their "air". One can look at ancient UI paradigms, specifically SunView, OpenLook and Motif (I'm currently excited about Sun history again), Windows 3.*, and also Win9x (with WinXP being more or less inside the same paradigm). And one can see that of these only Motif had anything resembling their "air". And Motif is generally considered clunky and less usable than the rest of the mentioned (I personally consider OpenLook the best), but compared to modern UIs even Motif does that "air" part the way it seems to make some sense, and feels less clunky, making me wonder how is that even possible.

FFS, modern UI designers don't even think it's necessary to clearly and consistently separate buttons and links from text.

And also - freedom in Web and UI design has proven to be a mistake. UIs should be native. Web browsers should display pages adaptively (we have such and such blocks of text and such and such links), their appearance should be decided on the client and be native too, except pictures. Gemini is the right way to go for the Web.