Main article: brain-computer interface.
How hard could it be? You just have to learn the encoding of the neural spine/eyes/ear, add an invasive device that multiplexes it, and then the benefits could be mind blowing.
Interestingly and obviously, the initial advances in the area are happening for people that have hearing or vision difficulties. Since they already have a deficient sense, you don't lose that much by a failed attempt.
Hearing is likely to be the first since it feels the simplest. Ciro heard there are even already clinical applications there. TODO source.