I'd be interested to see an independent audit on how much of this data they (and google/facebook/apple/etc) are collecting is being used to help improve training of Cortana as an assitant (or other tools that directly assist the user like Google Maps) , and how much of this data is really just for selling to advertising companies, and how much is mandated by governements.

I know this isn't binary some data may be dual purposed (training and ads for example) but I'd like to see an audit of what it's used for.

I'll never get that, but it would be interesting.