About this project
During lockdown, Jesse and Robin dropped cassette tapes on each other’s doorsteps in Oakland, a ping-ponging analog process that began as an open-ended exploration and sharpened into a coherent set of songs.
The rules for the process were simple but strict:
- We traded only mono cassettes; no digital files, no documents, no other information or intention. The choice of what material to use —
and what not to — was entirely up to each tape’s recipient.
- When Robin received a tape from Jesse, full of interesting musical fragments, he selected his favorites and used them to prompt an AI trained on decades of recorded music. Robin also provided the text of lyrics, guiding the machine as it generated responses that were by turns strange and surprising, frustrating and beautiful.
- When Jesse received a tape from Robin, full of these AI responses, he raised traditional musical scaffolding around the material, working both with and against it; making nice furniture out of funky driftwood.
Here’s what we think is interesting about the result:
- The process was totally organic, each of us working only with what we heard on the tape we’d been given. Even so, coherent themes emerged, which you can hear on the album.
- Two flavors of chance worked together: first, the fact that we didn’t know what we’d receive from each other; second, the fact that we didn’t know —
could never totally control — what the AI would generate.
- The vocal performances on this album are the work of the AI, but they don’t sound robotic at all. In fact, they’re the opposite of auto-tuned digital perfection: raw, noisy, melancholy, a little fucked-up. They offer a real sense of ghostly connection.
- This album is one of very few fully-developed music projects to use an AI model in this way. We hope it represents a step on the path from technical tool to expressive instrument —
the same path taken, in decades past, by the piano, the electric guitar, and the synthesizer.
In a sense, this project was our expedition to an uncharted Shadow Planet, and these songs are what we brought back.
In this interview with Clive Thompson, you can totally detect the rhythm of our live conversation, all three of us —
In this Q&A with James Vincent at The Verge, Robin gets into the feeling of using the AI model: “wandering in an enormous labyrinth or a dead city.”
Jesse Solomon Clark is a composer for emergent and traditional narrative forms. He has created music for films, commercials, and physical installations around the world.
Robin Sloan is a New York Times best-selling author of novels published in the U.S., Japan, and many other countries, and a creative technologist with a focus on AI.