I wonder if, within my lifetime, it will be possible to tell a digital assistant to "order all the ingredients I need to bake the cinnamon buns as detailed in the cinnamon bun baking video I watched last month with the Chinese lady that doesn't have an accent"?
It'll be useful when you can have it create, say, a menu and shopping list for your week, interact conversationally to tweak it including understanding that something like "I'm allergic to that ingredient" means never ever try to include it again while other statements of preference may be much softer, learn your preferences so those conversations get shorter every week, and order the ingredients for you, too. It barely matters if that happens over keyboard input or voice, because the hard parts of that problem have nothing to do with how the input is done.
You want to be able to give a few parameters for a trip you're taking and get some choices back. But, as you say, that's far more about "intelligence" than the type of input.
Nothing reminds you about the difficulty in cleanly parsing human language more than trying to work through a voice interface. I'm not sure what is missing precisely, but I'd say "an ability to use semantic understanding" might be helpful. All of our AI works explicitly with no understanding of semantics, but it's becoming clear to me that the ceiling that places on their "intelligence" puts them still leagues away from the world of true AGI.