Our consciousness is single-tasking, so to mullti-task we have to timeslice, and the particular problem there is that context switching, for us, takes more time than whatever task slice we are switching to or from.
So far as a motor vehicle is concerned, this is why learners tend to be planks. They know how to drive but have to think about it, so it's slow and consumes attention better spent looking for hazards. Once the driver drives enough, the act of driving is automatic so no longer requires a conscious task.
This is why touch controls are so shit at the moment - with physical buttons and stuff we learn where they are and how they feel, and eventually they disappear from our conscious thought (but not completely - that's a diversion for another time, perhaps). With touch controls there is no feedback so we have to look and search for indications that something has happened (correctly, mostly). That's conscious effort which detracts from driving. However, as we get used to them, perhaps touch controls will eventually merge into the background. It might need some additional help - perhaps a voice prompt - but once we are used to a repeatable means of operating them, and can then forget about the mechanics of doing so, they will be fine.