IP and entertainment lawyer Ash Kernen has been talking to Mega Labs about his views on generative AI and music.
“I think that ultimately we are going to have a bifurcated media content world. At one end of the spectrum, you’ll have the purists. The human-driven, qualitative, personality-ascribed media world that is very much the sort of farm-to-table movement of music,” he said.
“At the other end of the spectrum, you’re going to have the purely AI-driven McDonalds of music where it’s sort of mass-commoditised and middle-of-the-aisle. And then, there will be all kinds of variations in between.”
Kernen predicted that a new generation of “sort of prompt engineers, ‘prompt superstars'” will emerge as part of this in music.
“But here’s the thing… the average listener, they just don’t care how music is made. All they care about is the consumptive value of music. So if it slaps, it fucking slaps.”
Kernen also suggested that there are loopholes in many existing contracts between artists and labels where the labels “have no real claim to the impression of that artist’s voice. They have a claim to the actual use of that artist’s voice, but AI modeling is certainly not that”.
He predicted that labels may start to expand the terms of new deals to include the artist’s vocal likeness as well as their actual recordings.
“So it’s interesting because it’s really going to be up to the artist to align with the label to enforce those rights. But right now, a smart and savvy artist who’s unrecouped, might be like, ‘Fuck you, I’m licensing my voice here. Not you. Tell me where I can’t do this under my agreement,’ you know?”
“Savvy artists may actually lean into this. They’ll train a contained language model on their own catalogue, and then sell that as a plug-in…”