Just because you can bootstrap a computer to run arbitrary programs doesn't mean you're significantly closer to general AI.

Similarly, it's easy to argue that there will be incremental improvements in space technology, but it's very hard to argue that they will snowball into large-scale luxury platforms that can sustain human life indefinitely.

On the contrary it's very easy to argue it. All you have to do is type some words on a keyboard. Actually doing it however...

The earth has a wide variety of resources important to comfortable human life. The moon has rocks and sand and sunshine. Moon mining is but one of the prerequisites to living cheaply in space.

Yes. We'll probably move into space only after we are good into autonomously turning water, organics and some rocks into comfortable habitat. Not earlier.

That is probably not too far away. We are getting very good into semi-autonomous transformations, and at turning anything into the materials we want.

If this were true, one would expect life on earth to differ dramatically from what one observes today. A lack of necessary materials keeps billions of people in poverty.

Also because we're propelling it into space using disintegrating totem poles. Lofting it using something like a momentum exchange tether, launch loop, or space fountain would also make habitats of that scale much cheaper. Better materials science could also significantly reduce the necessary lift mass; inflatable shelters, for example, scale really well as they get bigger.