At the end of the day the benefits have to outweigh the privacy risks for people to not reject it. That’s why it’s baby steps to get people to want and accept AR . Why do you think I’ve been doing a furniture shopping app for the last 4 years; because I love furniture? Please. It’s to get people comfortable with the concept of AR and contextual visual computing and more advances human machine interfaces. People already share way more data than ever because they get more value out of it than they lose.
Lets just look at the problem technically for a moment.
Fast forward into the future 20 years and we have virtual retinal display systems which have superseded the smartphone, tablet et al. as our primary means of content creation/consumption.
One of the promises of AR is that it’s contextually aware, so that means it’s like a little human (or assistant) sitting on your shoulder at all times looking for context and giving you relevant information. Just like our phones now, you want to be able to use this system in public and private places and it should be seamless. So if I have content in my refrigerator, like expiration dates and nutrition info, it should be persistent when I open and close the door. Same for my laundry room, bathroom etc…
I should be able to seamlessly walk out of my kitchen and look down the street and call up an estimate on the time my mail will arrive because the global AR system is updating itself based on the input of the postal carrier.
You can’t do all of these things seamlessly with local queries because they are not updated with global optimization. In fact it would be totally counter everything we’re doing in AI to limit where information/data resides, as the collection and analysis of this data is critical to forecasting and improving decision making support.
I think there are a few things in here that are worth explaining further but I’ll leave it with this: My goal is to build a consistently updating decision optimization system with a human interface as the training and action mechanism — basically the largest Reinforcement Learning system ever made which helps people make better, data driven decisions daily.
That means we need to build relationships between human goals and rewards through some kind of system — I think AR is probably the best system for that as it’s persistent and lends itself to collecting both revealed and stated preferences rather than simply stated preferences like systems do now.