I did a project once for a touchless interface using a Leap Motion (remember those?) for use in operating theatres. The software was a 3D viewer for MRI scans. The surgeons attached to the project were excited, because currently they operated the software by telling the nurses how to rotate and zoom, which didn't work so great.

In the end it ended up not working well enough (the Leap Motion would often give glitchy positioning data), but I believe it's the way forward.

Yeah, those have been around for decades on all kinds of old specialised keyboards and equipment... The real reason stuff trends towards touch screens is people trying to keep the BOM cost low without considering cost to the user.

A generic touch screen with dynamic controls is cheaper than a larger custom made keyboard or physical controls. It's the same reason we end up with power inefficient IoT devices that could be made with very cheap and efficient microcontrollers, but developing for those requires expensive expertise, so we end up with huge powerful ARM cores because it's more general and easier to code.

The common factor in both of these issues is cost to manufacturer vs cost to user (power, safety etc).