Touchscreens are ideal for consumer devices because it means not having to commit to a single UI design - you can move buttons around onscreen as needed, overhaul the software to put things in a new place as you like, and not having to worry that the dial the user turns is tied to a value box that is now on the opposite side of the screen. Not only is it cheaper from a component perspective, but it's cheaper from a design perspective - you don't have to finalise the design for the UI before it goes to manufacturing, just tell the factory to stick a large enough touchscreen in the centre and the UI can be added later, after the various teams have finished arguing.

What they are not, however, is intuitive, much as designers like to say they are - there is zero consistency in the UX between devices, buttons all look different, sometimes between updates, and half the time you open an app, it's a dice roll whether it's going to work the same it did last time. There is also no inherent tactile feedback - phones have haptic vibrators to simulate a button press, but there is no other feedback, and you can't feel the buttons with your fingertips before pushing them. You have to be looking directly at the screen to find the control you need - when the device is primarily designed to be in your hand, it's kind of acceptable, but when it's a control panel for something vital, this simply won't work; your eyes are supposed to be fixed on what the control panel is controlling. These reasons are why Mazda has backed out of using them for cars - your eyes need to be on the road. You can change fan and temperature settings by touch alone once you know where the controls are. In Apollo 13, if Jim Lovell had to keep looking at a touchscreen to make sure he was actually pushing the thrust control in the right direction while keeping his visual fix on the Moon, there's a great chance he'd have sent the stack spinning off into space.