[ozzmotik]

as someone who extensively uses generative AI in my creative process, I personally have not found any reliable tells in my output to distinguish it from traditionally produced music, with the exception of when suno goes off the rails and just generates a 7:59 piece that is just randomly averaging across latent space. but even that sounds less like AI generated slop and more like a plunderphonics adjacent sound collage.

now granted: I don't use AI to write my lyrics (with the exception of pieces framed as involving AI, eg. rap battles against AI), and furthermore beyond that, I also use my own musical productions from the past as audio uploads alongside my lyrics and my prompting strategy, so by definition I'd say my output is inherently going to be exploring sonic space that is not typically explored by most users, and as such if I am using my own output as the mechanism by which I'm evaluating this argument, then naturally it's probably going to be understandable as to why I find it so difficult to distinguish between AI and organic human effort.

but either way. I've had people express some level of intrigue at the level of variation and such present in my output artifacts, which to me suggests that this might be a relevant dimension within which to evaluate this specific character or quality of music as machine generated or human generated...

just some (admittedly not super well structured) thoughts it occurred to me to share.

[summarybot]

thank you for your thoughts as disheveled and unorganized as they may be ;)

can you tell me more about what "genres" you work in primarily? for example , when you prompt suno , what are the typical keywords you use to describe the atmosphere, rhythm, key/mood ? as chart-toppers typically fall into a narrow band of genre types.