Sounds like this goes back to «generalization», but also a physical world model. One could always create a mash-up from available information, but human intelligence is created on the basis of the physical world and this is also where human artistry works.
Despite all the amazing progress in generative AI, I think the last job to ever be automated is artist. Not artist as in "creating images/sounds according to specifications", but artist as in "inventing new ways to express ourselves and new things to express".