"neural-computer interfaces" are further off than you think. Even if you do decide to stick electrodes where electrodes don't belong (and you really shouldn't do that), finding the signals for fine motor control and then controlling for all the spurious stuff that isn't fine motor control is a hell of a lot more effort than, say, a mouse.

I talked to the main brain imaging guy at the UW Dept of bioengineering back in 2005 to research a screenplay. One of those guys that was imaging thought using fMRI and the like back before it was cool. There's a whole bunch of "yeah, in theory we can do this, as soon as we solve this, that and the other thing, which we have no idea how to do, but we can see the pathway." I heard similar verbiage from Geoff Landiswhen I was consulting with him on crossing space-time using Visser wormholes. "Yeah, in theory we could open a gateway between two points as soon as we figure out how to harness 100% of the energy of a red dwarf star and if the reference points of the two ends are completely at rest because otherwise we have to absorb energy produced in accordance with Einstein's equation of general relativity."

We use our fingers to do things. They've evolved for it.

You might enjoy reading this. It illuminated a lot of the misgivings I had about UI and ergonomics that I hadn't been able to put into words.