One argument against AI labelling is that separating AI from “real” is complex. If artists use AI to fill a drum track, but they composed and performed the music themselves, should that be considered AI? Perhaps not? But where that line is drawn is absolutely open to debate.
Some highly respected figures in music see AI as a great opportunity. Hip-hop icon Dr Dre called it “a new tool for creativity”. When I discussed AI with ABBA’s Bjorn Ulvaeus for this column last year, he said AI helped him quickly imagine how it would sound when vast orchestras were playing a new composition. “It’s like an extension of my mind,” he said.
Still, even with that nuance, I believe that music generated entirely by AI, created with prompts rather than talent, is an abomination – and finding it placed by Spotify’s algorithms into a playlist is, as I wrote, a “deeply violating experience”.
ENCOURAGING DEVELOPMENTS
I’m pleased to report there have been some encouraging developments since. Earlier this summer, music streaming platform Tidal announced that under its new AI policy, AI tracks could be put on the platform but that the use of AI would be clearly labelled and, here’s the best bit: AI-made tracks could not be monetised.
Tidal is drawing the line at music that is “wholly or substantially” generated by AI. The company has said it will keep revisiting its policy as the technology evolves.
Tidal’s approach is the strictest so far, but other platforms are taking important steps, too. Starting in mid-September, Spotify has said AI-generated “personas” will be labelled clearly and excluded from appearing in algorithmic playlists.
Apple has pledged to do the same, promising to clearly label AI-made music on Apple Music. These measures follow the pleas of influential music groups, including the International Federation of the Phonographic Industry (IFPI), which in July issued a joint proposal calling for a clear and standardised labelling system for AI-generated music.
