I agree with your philosophical stance, in general, but this particular use case is one that I've been wanting for years and where I think altering the image can be in some ways more "honest" than showing the raw camera feed.
With an unfiltered camera, it looks like I'm making eye contact with you when I'm actually looking directly at my camera, and likewise it looks like I'm staring off to the side when I'm looking directly at your image in my screen.
A camera centered behind my screen might be marginally better in that regard, but it still wouldn't look quite right.
What I'd really like to see is a filter for video conferencing that is aware of the position of your image on my screen, and modifies the angle of my face and eyes to more closely match what you would actually see from that perspective (e.g. it would look like I'm making direct eye contact when I'm looking at/near the position of your eyes on my screen).
You could imagine this working even for multiple users, where I might be paying attention to one participant or another, and each of their views of me would be updated so that the one I'm paying attention to can tell I'm looking directly at them, and the others know I'm not looking directly at them in that moment.
Would be funny if everyone on your screen gave a side eye to the bottom right corner where the currently speaking person is.
Jokes aside, I think you're absolutely right. Online interactions have dynamic geometry, so mounting a camera behind a screen will just not cut it, unless the entire screen is a camera. Also, some people might prefer projecting/receiving no eye contact at all, at times, in situations. And vice versa.
Philosophical stance here is purely traditionalist, it decides on behalf of people. What people would like to use, that should exist. "Videos are and will" is a strange claim, assuming its claimer has neither control over it nor any sort of affirmation that it is going to be true.
Once we have technology to put a camera under a screen without sacrificing display quality ... we will not stop at one camera.
There will be an array of cameras covering say every 2x2 inch square of your screen.
Just see how many cameras are on todays phones. Same can happen with new camera tech too.
Also there will be a huge commercial driver to put multiple cameras under the screen -- all apps and marketers can track your precise gaze. Ads will pause unless you are actually watching them. I will hate it but it feels inevitable.
If that happened I would become a drastically different person because no one may control people’s bodies like that except for themselves. God that really made me angry to read. I really really hope you are wrong.
Ads already pause if you switch apps on mobile, and vending machines/retail screens have had cameras in them for expression/attention tracking for years. It’s not much of a leap from there
“Eye contact” is not a monolith though. Typically we look at someone’s eyes when we are speaking but their mouth when they are speaking. And eye contact can be a pattern of crossing between their left and right eyes. And making and breaking eye contact are important parts of nonverbal communication. The typical AI “eye contact correction” will do none of this.
It's also extremely culturally dependent. (Never mind that plenty of people in countries that obsess over eye contact find it uncomfortable as well.)
It's generally considered rude or an act of intimidation to maintain eye contact with people in Japan, for example. Not nodding occasionally while someone is talking is also seen as a sign that you're not paying attention. Are we going to modify videos to nod automatically too? Or maybe we can stop trying to fake social interactions and enforcing local customs on the world.
With an unfiltered camera, it looks like I'm making eye contact with you when I'm actually looking directly at my camera, and likewise it looks like I'm staring off to the side when I'm looking directly at your image in my screen.
A camera centered behind my screen might be marginally better in that regard, but it still wouldn't look quite right.
What I'd really like to see is a filter for video conferencing that is aware of the position of your image on my screen, and modifies the angle of my face and eyes to more closely match what you would actually see from that perspective (e.g. it would look like I'm making direct eye contact when I'm looking at/near the position of your eyes on my screen).
You could imagine this working even for multiple users, where I might be paying attention to one participant or another, and each of their views of me would be updated so that the one I'm paying attention to can tell I'm looking directly at them, and the others know I'm not looking directly at them in that moment.