AI Music Programmers, Machine Learning, New Genres Created by Machines, and the Radical Claim That Humans Are Optional in Music
CJ Carr and Zack Zukowski are Dadabots -- the musician-programmer duo who have spent years building AI systems that generate music autonomously, in real time, without any human playing a note. Their work sits at the sharpest edge of a debate that the music industry can no longer avoid: what happens when the machines get good enough that the question of whether a human made it becomes genuinely hard to answer? Their AI has generated metal, jazz, and entirely new genres that don't have names yet. They've livestreamed AI-generated music for weeks at a time, nonstop, without a human touching an instrument. And they don't apologize for any of it.
In this conversation with Elmo, CJ and Zack go deep on the technical and philosophical dimensions of what they're building: how their neural networks learn to generate music, what "eliminating humans from music" actually means and why they stand behind the provocation, the new sonic territories AI has opened up that human musicians simply can't access, and what they believe the music industry gets fundamentally wrong about the threat and the opportunity that AI represents. For musicians trying to understand what's coming, this is the most honest conversation available.
"The AI isn't trying to replace you. It doesn't know you exist. That's the part that should make musicians think harder."
The technical architecture behind their systems: how neural networks learn musical patterns from training data, what it means for a machine to "understand" music at a structural level, and the specific engineering decisions that separate music that sounds authentically generated from music that sounds like a bad imitation of something a human already made.
The provocation that defines Dadabots and what they actually mean by it: the philosophical argument behind the claim, why they believe it's more honest than the "AI as tool" framing most technologists use, and what the music industry consistently misunderstands about the nature of what AI can and cannot do at the frontier of generative music.
The music that emerged from their systems that doesn't fit any existing category: what happens when a neural network processes extreme metal or jazz at scale and begins producing outputs that have internal logic but no human referent, and why CJ and Zack believe these new sonic territories are genuinely exciting rather than alarming.
The technical and artistic achievement of livestreaming AI-generated music continuously for weeks at a time: what their systems are actually doing in real time, the difference between pre-generated AI music and truly live algorithmic generation, and what audiences responded to when they tuned in and realized no human was behind what they were hearing.
Their unfiltered assessment of how the music business is responding to AI: the arguments that frustrate them, the conversations that give them hope, and their honest prediction of where this technology is going regardless of whether record labels and publishers decide they're comfortable with it or not.
The dual identity of Dadabots as both musicians and programmers: how musical training shaped their approach to building AI systems, what they hear differently because of their backgrounds, and why they believe the most interesting work in this space will always come from people who understand music from the inside rather than people who only understand the technology.
CJ and Zack on what "eliminating humans from music" actually means: the full philosophical argument, why they chose the provocative framing, and why they believe the watered-down "AI as creative tool" narrative is less honest about what the technology is actually doing and where it's actually going.
The new genres their AI created that don't have names: specific examples of sonic territory their systems produced that no human musician was aiming for, what those sounds actually feel like to listen to, and why they believe the most interesting output isn't the AI that sounds most like a human -- it's the AI that sounds like nothing a human would have made.
On the music industry's response to AI: the arguments they find intellectually lazy, the protections they think are reasonable and the ones that aren't, and their honest view of what the next five years look like for professional musicians navigating a landscape that is changing faster than most industry bodies are willing to admit.
The technical reality of real-time generation: what their systems are actually computing when they stream music live, the difference between a pre-trained model playing back and a system genuinely generating in the moment, and why the distinction matters for how musicians and listeners should think about what they're experiencing.
Their advice for musicians right now: what they think working musicians should actually do in response to AI, the opportunities they believe most people are missing, and what they want the musicians listening to this conversation to understand about the technology that the loudest voices in the debate consistently get wrong.
How musical training changed the way they build AI: the specific things they hear in machine-generated music that pure engineers miss, how their backgrounds as musicians shaped their technical decisions, and why the best AI music tools are going to come from people who can hear the difference between music that is structurally correct and music that actually feels like something.