a)
Internally, MS staff also refer to Cortana as 'she'. As a result, so do many technical publications and that filters down to users as well. Despite all that, I don't think there is any real danger of anyone thinking there is a little woman hidden away in their smartphone.
The idea is to simulate, as closely as possible, a human personal assistant. That is the goal MS aspires to, and in that context it's reasonable to refer to that piece of software as 'her' (or optionally 'him'). It's partly about making it clear to everyone involved what the goal is. I don't think that is a bad thing.
b)
Nothing, absolutely nothing, about artificial intelligence or voice recognition is basic. Seriously, try to write down the math that can differentiate human voices from ambient noise and recognize individual words, and then watch how fast you fail.
Once you've got that, then realize that humans have dozens of ways to say the same thing, so just recognizing words isn't enough, you must also understand sentence structures and computationally be able to derive meaning and intent. And then do that for 20 different languages with different grammar and phonetics.
If you're talking about just a single language, then that is something that is trivial for a human brain, but it's not at all trivial for what is essentially nothing but a calculator.