I was asked by PRS today my thoughts on AI being used in music and if I was worried. I thought I’d publish my response here if it is any use to your thoughts. Are you worried?
In the realm of the arts I will only see AI as a tool, just like a matrix sequencer for a synth. To speed up my workflow will be great but client expectations are a concern, I can hear them say “you can’t produce this track in a day?!”.
In regards as an artist and how a consumer interacts, I think people will always be interested in people. For an AI to compete with all the jobs of an artist and connect with an audience seems like a stretch without a huge team of people behind it, physically presenting and maintaining the narrative. Would The Gorillaz be as cool without knowing the people behind it? Will Banksy turning out to be a machine devalue the work? AI commenting on social issues would be interesting but with no accountability?
Working in production music is tough already, with generally low pay for hours of work for big television shows and Indie composers up against monopolized and sewn up deals. I think if an editor had the function to “populate edit with music using these meta tags” and it was as good as Hans Zimmer then sync licensing for underscores would be threatened, with the odd novelty human contributing now and again. From my experience working in the advertising industry I know they currently tend to not favour AI as it is unpredictable and they often need someone on hand to fix things immediately or develop something organically but when we cross the threshold to when an AI can be directed with accuracy then I imagine I’ll consider becoming a farmer and to be honest this job is tough enough that I consider that without AI in the picture. One last thing to mention on the already tough landscape composers and artists are in – quality does always rise to the surface and we do need to protect that quality from being mass-produced without our consent or rights infringed. I do feel like music has been devalued and popular tastes have narrowed over the past 40 years so I anticipate AI being able to fill that vacuum and produce Tik Tok music well but there will always be a need for a live human performance and tempered compositional choices.
Should we take further steps to copyright that performance or “our sound” or tonality and timbre of our voice or instrument or writing style to protect it from being used in datasets? A work around could be anything released for profit by an AI will need to fully disclose the dataset, akin to disclosing the samples you’ve used in your hip-hop track and a team or another AI (IKR) works out if it should be demonetised.
I’m truly fascinated if AI will produce something accessible and new and not just be a freaky cookie cutter. Having our rights intact and our music valued is an ongoing battle and this will be another player among many to pitch against. Would I… use algorithmic engineering to give starting points for a syncopated pattern for 15 instruments – yes, generate lyrics with chatGPT – no, use an imitation instrumentalist to perform parts – yes, I already do using virtual instruments.