"Machine Learning" isn't in its infancy yet. It hasn't even been physically conceived yet. It's an idea so far.
And we don't have artificial intelligence, we don't even have artificial stupidity. We have a parser that, in this case, thinks it "understands" what you want before you finish and a fair database lookup.
Android Assistant isn't much better - a lot of questions are "answered" by running a Google search of the words you spoke and presenting you with the Google page - in effect "I have no idea".
Some day we might have artificial stupidity, but I doubt that anyone who's been born yet will see it. Artificial intelligence? Maybe in 5 centuries or so. You note down a doctor's appointment. The phone app knows that you like to get to that doctor 10 minutes early. (It keeps track of things like that without being asked to or being set up to - the way a person would.) It starts checking Maps about an hour before you'd normally have to leave, to make sure that, if theree's a major accident, trying things up for many blocks, you can still get there 10 minutes early by using an alternate route it plots for you. Then, 5 minutes before you have to leave it lets you know that you'll have to be leaving for the doctor in 5 minutes. All you entered was Dr. Jones, 2:30 PM, September 20. The AI in the phone took the initiative to take it from there. Maybe in a few centuries.