That being said I think it misses what made the old physical interfaces so appealing and useful. It's not that there's something inherently superior about multimodality; it's that physical interfaces are permanent, with defined edges and definite shape. Unlike screens you know exactly what's where, building muscle memory every time you use it. There are no hidden menus or moving parts.
Multimodality - such as being able to see the position of a slider at a glance, or feel its position by touch - is useful because it reinforces the absolute existence of a control and its state across multiple senses. Interfaces using voice and gestures like suggested are the exact opposite of that, because each point of interaction becomes even more disconnected and vague.