Open Source: Reclaiming Generative Music

When I first got involved in generative music, the idea of it seemed mysterious, esoteric, and alluring: the example of artists like Brian Eno, Laurie Spiegel, and Autechre using technology to create semi-autonomous systems that explore new musical worlds was, and still is, a north star in my work.

However, in the past couple of years, the terms "generative" and "generative music" have come to mean something altogether more banal in the public imagination. When you hear about generative music now, it's usually in the context of the latest GenAI startup flooding social media with sterile prompt music gunk, based on datasets with spurious provenance claims. A synonym for musical factory farming.

And yet to take a simplistic "anti-AI" stance seems silly. It downplays all the genuinely interesting exploration people are doing in the generative machine learning space. It leaves on the table entirely new ways for artists to create. It also just feels like a slide into an old-man-yells-at-cloud mindset I'm not comfortable with.

So, what to do? How can we reclaim generative music and make it cool again? How do we incorporate and amplify all the good stuff in the AI space and resist simple binary oppositions? I'm not sure I have the answers to these questions but I'm keen to explore them.