AI Slop Exists. We Are Not It. Here's the Proof
The paper informs, and the music speaks for itself.
There's a conversation happening in music right now that needs to end. It goes like this: someone hears an AI-generated track, it's mediocre, and they conclude that all AI music is mediocre. That all of it is slop.
They’re not wrong that slop exists. They’re wrong that slop is all there is.
This is the full 82 minute Indian Classical / Dark Electronica audio file, download for free. It needs to be heard.
Last night I published a peer-reviewed preprint on Zenodo that makes this argument with evidence. It’s called Prompt as Score: Compositional Literacy in Human-AI Music Generation, and it was co-authored with Claude (Anthropic). It’s on the Bitcoin blockchain. It has a cryptographic fingerprint. Anyone can verify it exists and who made it.
Here’s the core argument: the gap between slop and craft in generative AI music is not a model problem. It’s a literacy problem. A generative music prompt written by someone with deep musical knowledge functions as a complete compositional score. The model reads it and performs it. Hand that same model a low-information prompt and you get the most probable output — undifferentiated, generic, forgettable.
AI slop exists. But AI slop doesn’t write academic papers about what it is. It doesn’t hash its outputs to the Bitcoin blockchain. It doesn’t name its contributors, credit its AI collaborators, or publish a Code of Conduct as its product interface. It doesn’t have a 35-year lineage running from co-founding the music element of Burning Man through working sessions with Timothy Leary, Robert Anton Wilson, and Genesis P-Orridge to a peer-reviewed preprint published at 5am.
Slop is anonymous by nature. Nobody signs it, because nobody wants to be accountable for it.
We sign everything.
The Zenodo paper documents a single production session that produced Om Namah Shivaya — an extended devotional work built on a prompt encoding G Phrygian mode, 68 BPM resting heart rate entrainment, solfeggio frequencies, a full hypnotherapy arc, and a guitar player on a heroic dose of LSD who effing nails his solo. Every element deliberate. Every choice documented. The result: music the practitioner described as the most beautiful he’d ever heard, consistently, across every generation.
That’s not luck. That’s 35 years of accumulated knowledge meeting an instrument capable of responding to it
.
The paper is open access. The music is on the record. The provenance is verifiable by anyone at verify.haawke.com.
We are not slop. We never were. Now it’s on the permanent record.
Read the paper and listen to the music: https://zenodo.org/records/21456110
After listening to the music, come back and tell me what you heard.
Pictured - above: Craig Ellenwood (AI Enhanced) Bottom: Claude (ai generated)
#AIMusic #GenerativeAI #HumanAICollaboration #MusicProduction #ArtificialIntelligence #Ethics #Provenance #Blockchain #CreativeTechnology #MusicIndustry #Innovation #ThoughtLeadership #Claude #Squaawke