As a songwriter and musician who has often played in support of other song writers, I believe (generally) everything we enjoy has been reworked from some prior existing work.
One issue I’ve seen, especially in younger songwriters, poets, and musicians is that they are less aware about where they have sourced the material which they are reworking.
This lack of awareness is generally a-okay: you can get a lot of mileage reworking quotes that you are not aware are from the christian bible.
Compelling work is compelling, after all.
However, what I personally find artistically satisfying is when people take multiple sources which resonate with one another and which, in their harmony, create rather complex symbolic patterns.
At the micro level, that’s just rhyming words; at the macro level that would be thematic development.
Younger, less-well-versed writers miss a lot of opportunities to tie what they are creating to other texts because they are simply not familiar with other ideas or existing works. This doesn’t mean that they are creating bad art, in my understanding; rather, they they could do a better job of their task of stitching words into some meaning.
In the context of LLMs, we have tools for stumbling around and creating those connections for us. I am sympathetic to the idea that there could be emergent connections that only a machine, via modeling language, could discover.
I don’t know so much about the machines’ understanding. What I do see has been folks who don’t care enough about whatever art form with which they are engaging to learn the source material from which they are drawing.
If that makes compelling work, then good on them.
At the same time, I am not sure how easy it is to create compelling work without that larger understanding of the context from which the work is dranw and into which that work is placed.
Or, if you prefer, there are a number of folks who have already summarized all my above thoughts as something like “great poets imitate and improve, whereas small ones steal and spoil.”