Creepy aside, this use case does sound, well, useful. But how often do we pass by someone wearing a pair of shoes we want?
Apparel, furniture, living rooms, bedrooms, office, exterior paint colors, cars you can't identify, etc.
The broader context is anything you can think of that is purchasable... we get our ideas about what we want by seeing or hearing them (I've used Shazam a LOT to tell me what a song is that I hear in a restaurant and then go buy it), and we often forget what they are or have no idea where to start looking. And cracking that nut solves a number of problems, including the marketplace for intrusive advertising or advertisers collecting data on you... if everything everywhere advertises itself, what use do merchandisers have for intrusive ads, or other middlemen, if software and hardware can connect you directly to the thing you want?
It's not going to entirely replace other ways of doing things, but it is a multi-trillion dollar market, something big enough that it's worth Apple's time and gargantuan cash reserve to go figure out... and it's a convenience that every single person who buys anything can use.
are people going to walk around wearing a pair of smart glasses that they most likely will have to charge daily or carry around a power source with
This is why I think this tech is about 50 years away. The firepower and the battery life tech aren't anywhere near what they'd need to be yet. But this is where I think MR would actually be useful on an iPhone-like level. Until then, it's a bit of a novelty that is more cumbersome than convenient.
EDIT: Also, this is just an example of the kinds of features that make this tech truly useful... productivity driven by wants more than needs, because desires drive more spending that does not need a business justification. There are other kinds of things we do that revolve around anticipating our needs, but that level of AI is even further off into the future.