My brand new (just arrived today!) stove has these capacitive buttons, and in terms of visual style it does indeed function perfectly; it's entirely clear which parts of the interface are buttons, even though most of them have only unadorned text.

But I think that analogy falls short in one major way, which is that these buttons are logically grouped and evenly spaced across a surface that consists of _nothing but_ buttons. If there's silkscreened text there, you can press it and do something. The visual display of information happens in an LED display set aside in a well-defined area, so things to look at and things to press are 100% clearly delineated. In a desktop/mobile/web app, where content and interactive elements are interspersed, you lose that obvious context clue. That inevitably adds cognitive load, even if the interface is still technically usable.