I think the actual answer is that you put yourself in the middle and then as you explore all around you you expand the map. So at the beginning you may choose South/North arbitrarily just based on the local features. Then as you explore you find out whether you are on top or bottom of the expanded map.

TBH this is one of the things someone knows the true answer and I could just Google it, but it's fun to speculate.

And since most land and therefore population is on Northern hemisphere, it was likely that whoever is in power will be there. And after all, it was a good choice as it's natural orientation for most people.

It's not 100% certain. There's a chance the Earth will live it's days without extinction level asteroid hitting it until the expanding Sun destroys it.

Depends on the medium, but human drawing is not that far from finding an image in a noise. If you start with empty paper, then refine it into sketch, then into colored image, then into more detailed image, etc. It's not that far off from refining image out of noise.

With img2img it is like drawing what it "sees", with text2img it's more like drawing something from memory.

Anyway, as I said most of the training data were photographs of the real world, those would still exist without human artists. I guess the pending question is if the AI could then develop some artistic style without inspiration from people.

My guess is that not at first, without non-realistic inputs it would only try to create realistic images. But there would be some distorted images and if any of them look good, they'll end up on the internet. Future AIs will be train on them as well and expand on them... It would eventually evolve into different artistic styles without humans, just take longer to get there. Good we had human artists to kickstart the process.

Well there's demoscene for example. I agree it's not mainstream, but surely not only mainstream art is the "real" art.

Sure, but you can say that about human artists too. "You could never draw this well if you couldn't see, you didn't make the things you see!" Most of the training data are photographs of the real world, some of it are creations of other artists.

It is strange to see this as a fight between good human artists and cold evil machines. The algorithm is a creation of many scientists and programmers who put lot of effort into it. I am amazed by it, because it shows HUMANS can create a thing that creates these pieces of art. I am looking forward to what more we can make.

I find it strange that people thought other places (like the heart) were where mind resides, because I can "feel" I am in my head, not other part of my body. Maybe it's just learned feeling from knowing the truth.

Also the placement of eyes and ears kinda place the center of self into the head.