Additionally, distractions are already abundant in a vehicle. A crappy user interface, which most are, is a safety concern. At minimum, lag between screens loading should be extremely minimal. Staring at a screen for an extra second or two while it registers that you pressed a button and updates the screen is incredibly dangerous and has surely lead to some number of deaths.
Two years ago Mazda announced they were moving away from touchscreens, which was very well received on HackerNews. [2]
And I can't find the thread right now, but 1-3 years ago there was a really good discussion on here about an eye tracking study in a variety of car models.
Touchscreen-only interfaces in cars are kind of like electronic-only voting machines. Technical people who know about computers breath funny thinking about it and see issues left and right while large swaths of people really want it.
The car producer could save even more money if they released the non-security critical user-facing parts of the software as open source and let car enthusiasts fix the bad user interface.
When purchasing complex goods like cars (or smartphones), there's way too many things to simultaneously optimize for, and enough confusion (mostly intentional) about relative performance, so most consumers focus on few major indicators like price, availability, cost of ownership, appearance, etc. Few people are going to trade on those major points to optimize something more specific, like lack of touchscreen (or having a headphone jack), so there's no meaningful market feedback on this, and vendors are free to dictate the choice to the market.