In my first car, I could operate the windows, climate control and sound system without taking my eyes off the road at all, although I had to glance briefly at the (fixed) display to see what radio station I was tuned to if it wasn’t obvious.
Old-school radios were a lot more user-friendly with preset station buttons and a tactile volume control that actually felt like it was connected to something.
With touchscreens, it's not just that you lose the tactile component, but all these interfaces are modal, with buttons that disappear or move around depending on the screen you're on.
Oh, you're on the radio screen? There's no way to adjust seat heating from here... or if there is one, it's in a different place than on the AC screen.
Those controls are typically on some surface of the car your hand is braced on. They're also very simple physical controls with a good amount of tactile feedback. It's hard to fuck up a simple push button window control or AC dial. Even on a bumpy road you'd be hard pressed to have trouble with such controls.
About 20 years ago, every teenager in the world who had a mobile phone was able to select a contact from their phonebook and type an entire message and send, in class with their phone in their pocket.
This is possible because of physical buttons and a deterministic user interface. The same applies to cars and other control interfaces.
I reject the idea of touch screens for car functions because it takes more than one level of navigation to reach the active button. If the UI designers would change the design so that when the car starts moving the touch screen would change and lock to a display where all driving-related buttons (heating/wipers/car stereo/gears) are close by, highly visible and activated with a single touch I could see myself using a touch screen in a car.
My main reservation is taking my eyes and attention of the road to focus on navigating a touch screen UI put together by 5 teams and 3 different committees.