Every time I encounter a conversation about artificial intelligence and independent music (and this is something that happens to me a lot), I notice the same thing. People use many of the same words, share many of the same concerns, but then often arrive at completely different conclusions, resulting in an impression of their counterpart as, at best, an idiot, and at worst, malevolent – not because they disagree, but because what they’re talking about are radically different concepts with very similar names. This is exactly the sort of thing that used to get people like me into trouble and onto combative conference panels about copyright.
On one side, there is what I would call AI Music: algorithmically generated sonic artefacts. These are the outputs of systems that produce audio, sometimes convincingly, sometimes not, with little or no human creative involvement in the compositional process. This is what most people picture when you say “AI and music” in the same sentence. It is what the headlines are about. It is what the lawsuits are about. It is, for many musicians, the source of a somewhat reasonable anxiety about the future of their livelihoods.
On the other side, there is Music AI: artificial intelligence technologies in the service of music creativity and music enterprise. These are tools that help musicians, labels, distributors, managers, promoters, choreographers, venues, writers (and, importantly, listeners – because let’s not forget that music is also communication and meaning-creation), and everyone else in the independent music ecosystem to do what they already do, but with new capabilities, new efficiencies, new modalities and new possibilities. Music AI is not about replacing musicians, though that is a genuine effect, and we do need to deal with that.
This is not a trivial distinction. It’s the distinction that determines whether you experience the next few years as something that happens to you or as something you have at least a degree of agency and deliberate control in.
I should make it clear that this is not a criticism of generative music. Algorithmic composition has been with us for a very long time, from Mozart’s dice games to Brian Eno’s generative systems to the procedural soundtracks of modern video games. There is a rich and legitimate creative tradition there, and it would be both inaccurate and unfair to dismiss it. The point is not that AI Music is bad. The point is that it is not the whole story, and treating it as if it were means missing the part most likely to be practically useful.
If you have ever read anything of mine before, you will know that I think of music as a fundamentally technological form. A goat skin stretched over a hollowed-out log to make a drum is a piece of technology. A musical scale is a technology. Notation, the concert hall, the vinyl record, the streaming platform: all technologies, all shaping what music is and what it can do. And importantly, we can not only use technologies, but we can also design them. We can hack into them. We can use them in ways that do not correspond with the manufacturer’s instructions. We can shape their meaning and purpose.
In 2014, at Music Tech Fest in Boston, we co-wrote a Manifesto for Music Technologists (the ‘Musictechifesto’) that put it this way:
“Music has always been technological. Bone flutes and drums are among the oldest known technologies.”
That manifesto asked what I think is still the most important question you can ask of any new music technology:
“For whom will this make things better? How?”
It argued that music technologies make worlds, and that we should endeavour to make better ones. That thinking still drives much of what I do.
It also means that none of this is particularly new to me, and I want to be clear about that, because the last thing the world needs is another commentator who discovered AI with ChatGPT in 2023.
In 2016, at Music Tech Fest in Berlin, we ran what we called Vocal AI Labs, experimenting with the intersections of artificial intelligence and the human voice in live performance. In 2019, we ran MTF Labs at the AI Impact Lab in Örebro, Sweden, focusing specifically on Music AI and Dance AI. Over the past decade, AI has been increasingly central to the work I have done and to how I think about the changing relationship between human creativity and technology. The reason I’m writing about it now is not that I have just arrived at the subject. It’s that the subject has arrived at a point where sitting down and carefully unpacking it has become genuinely urgent to an audience much wider than the leading-edge innovators and frontier artists I have the privilege of working with at MTF.
What I am interested in, and what I plan to write about here over the coming weeks and months, is the landscape of Music AI from the perspective of independent music, because that is what I am interested in. Not ‘how to’ use these tools. The world is full of YouTube tutorials and step-by-step guides, and I am not particularly interested in adding to that particular genre. I am not a ‘content creator’. But I do like to think about this stuff, and I sometimes like to talk about this stuff, and this seems a good space to do that. After all, it’s probably been sitting idle long enough.
What I’d like to try and map out is how to think about these tools: what their affordances are, what opportunities they present, what parameters are worth understanding, and what risks are worth taking seriously. As I’ve said for 20 years or more, in answer to the question “Should I be worried about this technology?”, the answer is always no. Worrying is not good for you, and it never actually helps. Understanding and acting, armed with knowledge, is a much better approach.
As with every significant shift in the technological environment around music, there are things that can help you and things that can harm you, and the difference is not always obvious. What matters is that you are equipped to make informed decisions rather than simply reacting to whatever the latest platform, product, or socially mediated moral panic has put in front of you. I am not here to tell you what to decide. I am here to lay out what there is to decide about. To help put the ‘think’ in whatever your ‘here’s what I think’ ends up being.
If you’ve been around New Music Strategies before, you will know that this is mostly what I do. When the internet was going to destroy the music industry, I wrote about what independent musicians could actually use it for. When social media was going to disintermediate and democratise, I wrote about what that actually meant in practice. Some of those observations turned out to be useful. I hope these will too.
For what it’s worth, I have what I consider to be the luxury of no longer being on social media, so people are unlikely to come across this in the usual sorts of ways people come across online articles. If you have found it, you read this far (thank you!), and you know someone who might find it helpful, please share it. Word of mouth is, as always, the most reliable and appreciated distribution mechanism in independent music.
And to answer your immediate question about whether I wrote this myself or used AI, the answer is yes. Yes, I wrote it myself, and yes, I used AI. As we go along, hopefully you’ll start to see how those two things can both be true, and how the more the latter, the more the former.