Amidst all the excitement around Siri, I came across an interesting article1 from the 90s, where Mark Weiser, ex-principal scientist at Xerox PARC, voices his skepticism about voice input and intelligent agents. His main argument is that interfaces are first and foremost tools, and as such, should be invisible.
Eyeglasses are a good tool—you look at the world, not the eyeglasses. […] Unfortunately, our common metaphors for computer interactions lead us away from the invisible tool, and towards making the tool the center of attention.
He goes on to explain that voice-based digital assistants, as portrayed in science-fiction, are prominent and attention-grabbing, which makes them inherently bad tools:
A computer that I must talk to, give commands to, or have a relationship with, is a computer that is too much the center of attention.
I agree to some extent; good interfaces should not vie for user attention. But then, the computer he imagined talking to is certainly not one you can fit in your pocket. In that sense, we are by all means moving one step closer towards invisibility.