Music generation, image generation all suffer the same way. You can make some beautiful stuff sometimes, but it (almost) always comes out 'hollow'.
I mean, are we pretending that this is a surprise?
I know we've spent the better part of the last century debating what "art" is, but I think we have fairly conclusive evidence that it isn't this.
No intention behind it, no human connection, none of the things we seek to find in art (including video games).
Everyone needs to read the famous essay 'The Work of Art in the Age of Mechanical Reproduction' by Walter Benjamin (1892-1940). He wrote about the Industrial Revolution, but his points become ever more salient in the age of AI.
His central thesis is the concept of "aura": the difference between a carefully crafted work of art and a mechanically mass-produced, but otherwise identical, object. When you describe something as "hollow", that's what's missing.
With music generation, it's good to not go all the way up the chain of abstraction.
Generate the synths, the DAW, the bass-line, the vocals, but never the whole thing. You can (and should) still compose by hand. The act of composing should be fun enough that you want to do it manually, otherwise you're not really the one making the song.
Also, a minimal amount of music theory knowledge helps significantly.