We have the same problem with key fobs in my wife's Escape. However I'm guessing it would be pretty easy to solve most cases, by setting a priority. I'm guessing most people are like my wife and I. If we are both in the car, 99% of the time a particular one of us is driving (in our case, me)
Yes, it's possible... this is "just" software - most anything is possible.
But it's going a rabbit hole of predictability.
How many seconds between "detections" is appropriate?
IE - let's say I'm set as the priority ...
It picks up my wife's phone first.
Then, 7 seconds later, I get in the car.
Does it leave the vehicle on my settings, as I'm the priority? Does it change to her profile because it didn't pick me up for 7 seconds? What's appropriate for the "priority detection timeout"? 5 seconds? 10? 2? anytime before the car's in drive? Do we make it user adjustable?
What if I'm standing in the driveway close enough so it detects me, but my wife's taking the car out somewhere? Should it switch to her settings if she opens the door and gets in but I don't? What if I get in a minute later? Would it switch back to me?
It's just a very complex set of scenarios, of which I can see many ways for the behavior to /appear/ unpredictable, when it's actually working as designed. Unpredictability is a user interface failure, no matter whether it's working as-designed or not.
But getting in the car and pressing "Here's who I am?" That's 100% predictable.
The more and more I think about it, the more and more the "leave the profile based on who drove last" makes sense.
In your particular case, since 99% of the time you're the one driving, 99% of the time it'll be on your setting anyway. Making it phone-detection-dependent would actually introduce unpredictability and increase the likelihood that you'd be on the wrong profile. That'd be counterproductive.... in your case, and most likely, in many cases. Mine included.
Now, if someone said "hey, let's use the interior facing camera for facial recognition of who's in the driver's seat...." -- we'd be on to something.