I’m imagining a spaceship landing at a deserted planet millions of years in the future and then just sitting there, generating pictures of cities in the sky and Mona Lisa on a surfboard.

Kind of a fun hybrid where AI is clearly dominant in an established solar system, but it ends up being impossible to miniaturize the enormous supply chain necessary to fabricate microprocessors, so to colonize a new solar system it actually ends up being necessary to send self-replicating carbon-based humans who will then bootstrap the industrial supply chain necessary for AGI.

(repeat ad infinitum for the rest of the galaxy, humans live only on the frontier)

It's all fun and games until you think about the practicalities of building a self-repairing machine that can keep running for hundreds/thousands/millions of years with no significant software or hardware errors.

It seems that there's a very pessimistic trend with regards to estimating the feasibility of engineering human biology for inhabiting outer space. Everyone just wants to let AI do it. I think this is because the progress of biotech has been so glacial. This is largely due to regulation. China will probably change the rules around at some point to allow for more flexibility here, with lots of moral outrage and handwringing in the west, and speed off into the future.