I actually suspect those aspects are among the easier ones to replicate. We have this idea of robots act as heavily autistic people. But when running Turing tests this often trips up people. Many times they think people who are not very chatty are computers, and mistakenly think chatty and funny programs are humans.

We have also seen that AI can generate beautiful looking art with things like Stable Diffusion, so I am not sure if aesthetics is a hard part.

In general it is a bit hard to assess as AI today trips up in ways which are very different from humans. You look at autonomous driving how it can avoid really tricky situations which a human could never avoid, only to do the most insane and obviously wrong move into another car or obstacle.

I suspect a computer doing humor would be a bit similar. You may hear a lot of clever jokes and insights only to on occasion hear stuff that is absolutely atrocious and offensive.