The AI and the DJ
Why the future of music depends on who controls the machine, not whether the machine exists.
The machine can generate a track. It can generate ten thousand tracks. It can generate a track that sounds, in isolation, like something a talented producer made on a good afternoon in a good studio.
What it cannot do — and what no one making confident predictions about AI and music is being careful enough to distinguish — is tell you which of those ten thousand tracks means something, to whom, in what context, under what conditions, at what moment in a night that has been building for three hours toward something the room does not yet know it is ready for.
That is not a small distinction. That is the whole thing.
The Machine Can Generate Output
Let me be clear about what AI can already do in music, because the honest version of this conversation requires facing it directly rather than retreating into anxious vagueness.
AI systems can now generate competent, stylistically coherent music across a wide range of genres. They can clone vocal timbres, reproduce production aesthetics, generate stems, master audio, write lyrics in the style of specific artists, and produce material that passes casual listening scrutiny at a level that would have seemed impossible five years ago.
AI can generate DJ promo mixes, complete with voiceovers in a synthetic version of a real person’s voice. It can produce plausible artist biographies and press materials for identities that do not exist. It can populate a social media presence, including images, captions, and engagement patterns, for artists who were never in a room.
None of this is speculation. All of it is happening, at increasing scale and decreasing cost, in the commercial music ecosystem right now.
The response to this from some quarters of the music industry has been panic, and from others has been enthusiasm, and both responses share the same underlying error: treating the output as the point.
Output Was Never the Whole Point
Here is what the ten essays before this one were, collectively, arguing: that music culture is not primarily about the production of audio content. It is about the production of meaning — through timing, context, community, memory, physical presence, taste, and the accumulated judgment of human beings who have lived inside a culture long enough to understand what it is for.
The track is not the work. The track is what the work produces when everything else is working.
The DJ is not a device for playing records. The DJ is a living system of musical memory, community context, physical attention, risk, taste, and decision-making — a system that produces a particular kind of collective experience that a playlist, however well-curated or algorithmically optimized, does not produce and cannot.
AI can generate output. It has never been in a room. It does not know what 4 AM feels like from the floor, or what a crowd that is starting to tire needs before it knows it needs it, or what the relationship is between the track currently playing and the eleven tracks that preceded it in this specific room with these specific people on this specific night. It has no body. It has no stake. It has no memory that is anything other than statistical.
These are not poetic observations. They are functional descriptions of what the craft actually is and what the technology actually lacks.
“AI can generate output. It has never been in a room.”
The DJ Is Not a Playlist
This distinction keeps getting blurred, and it is worth being precise about why it matters.
A playlist is a sequence of tracks organized by some principle — mood, tempo, genre, era, personal association. It is a useful thing. Streaming platforms have built significant commercial infrastructure around it. Algorithmic recommendation has genuinely changed how many people discover music, and in some respects for the better.
A DJ set is something structurally different. It is a real-time reading of a specific room — its energy, its history, its community, its collective state at this particular hour — and a response to that reading that unfolds as a continuous act of judgment, intuition, restraint, and risk. It requires the DJ to be wrong sometimes. To make a call that doesn’t land, to recover, to learn something about this particular room from the failure. It requires the DJ to be physically present, physiologically activated by the same environment the crowd is in, subject to the same variables of heat and sound and time.
You cannot replicate this with a system that has never been in a room, because the thing being done is not selection. It is listening, in real time, with a body and a history, to a collective that is also listening.
AI can produce a set. It cannot hold a room. The difference is not technical. It is ontological.
The Artist Is Not a Prompt
The same logic extends to music production, and the pressures are, in some respects, more immediate.
There are now platforms and workflows where a user can input a description — “deep house, melancholic, 124 BPM, influenced by early Detroit techno” — and receive a track that is sonically coherent with that description. The technology will continue to improve. The output will become increasingly difficult to distinguish from human-produced work on a superficial listening pass.
What this does not produce is what the previous essay called the living system of memory, taste, body, relationships, documentation, and cultural transmission that constitutes an artist. It produces output that resembles the surface of that system’s product. It does not produce the system.
The producer who spent twenty years developing a particular approach to sound — the specific tension between space and density, the harmonic decisions that are not formula but instinct built from thousands of hours of listening and iteration — is not interchangeable with a model trained on the outputs of producers like them. The model produces resemblance. The artist produces origin.
This matters for culture because origin matters. A music culture built entirely on resemblance — on outputs that sound like things rather than being things — is a culture that has solved the technical problem of music production while abandoning the human problem of meaning.
AI as Tool, Medium, and Threat
I use AI. I use it to organize, draft, research, iterate, and extend work across my projects. I am not writing this essay from a position of principled resistance to the technology.
What I am arguing is that the distinction between AI as tool and AI as author is not academic. It is the difference between a culture that uses new technology to deepen human creative work and a culture that uses new technology to replace the human investment in creative work because the investment is expensive and the replacement is cheap.
As a tool, AI is genuinely useful in music contexts. Organizing archives. Drafting press materials. Transcribing and analyzing recorded sets. Generating visual concepts for events. Handling administrative workflows that consume creative time without generating creative value. These are legitimate uses that support the artist’s infrastructure rather than replacing it.
As a medium — a new creative surface that artists work with intentionally, making deliberate choices about what the technology contributes and maintaining clear authorship over the result — AI has produced and will continue to produce genuinely interesting work. The question of what constitutes meaningful human contribution to an AI-assisted work is real and contested, but it is not unanswerable. The Recording Academy’s current position — that AI-assisted work can be eligible for recognition provided human creativity is meaningfully present — is an imperfect but honest attempt to hold that line.
As a threat, AI operates most dangerously not at the frontier of creative experimentation but in the middle of the market — in the economic layer where most working musicians, DJs, and producers actually earn their livelihoods. Session work, sync licensing, production for hire, the ambient commercial music economy: these are the areas where AI-generated output is most directly competitive with human creative labor, and where the humans involved have the fewest institutional protections.
The Likeness Problem
There is a dimension of the AI and music conversation that moves beyond economics into something closer to identity violation.
AI voice cloning can now produce synthetic versions of a real artist’s vocal timbre that are convincingly indistinguishable from the original in casual listening. This has been used to generate tracks, DJ promos, and promotional content featuring the synthetic likeness of real artists who did not consent to the use of their voice, their image, or their name.
This is not a gray area. It is the use of a person’s professional identity — the voice, the aesthetic, the cultural associations built over years of work and community investment — without consent, for commercial or reputational purposes the person has not approved.
The culture’s response to this should not be primarily legal, though legal frameworks matter and are developing. It should be ethical and communal: a refusal, by listeners, platforms, promoters, and industry participants, to treat synthetic likenesses as equivalent to the consent of the people they imitate. The market can only rationalize this if the culture allows it to.
The Memory Problem
Here is the subtler and in some ways more serious version of the threat.
Music culture depends on the transmission of memory — of what scenes felt like, how they developed, who built them, what the relationship was between specific communities and the music they created. This transmission happens primarily through human beings: through the artist who remembers, the promoter who carries the community history, the writer who preserves the record, the DJ whose set reflects twenty years of absorbed listening.
Get Saxe Coulson’s stories in your inbox
Join Medium for free to get updates from this writer.
AI does not remember in this sense. It has been trained on data. The difference between training data and cultural memory is not a technical distinction. Cultural memory is embodied, contextual, relational, and continuous with the living community that holds it. Training data is statistical, decontextualized, and frozen at the point of training.
A culture that outsources its memory to AI systems — that treats the statistical aggregate of its past outputs as equivalent to the living transmission of its history — will produce outputs that increasingly resemble its past while understanding it less and less.
This is not a prediction about a distant future. It is a description of a pressure that is already operating. The defense against it is not refusing the technology. It is investing more seriously than ever in the human systems that technology cannot replicate: the artist’s memory, the community’s archive, the relationship between generations of practitioners, the living transmission of what the culture is and where it came from.
✦ Human-Centered AI Music Checklist
A practical framework for artists, producers, DJs, and creative teams navigating AI tools with authorship and cultural integrity intact.
Authorship & Intentionality
- [ ] Be able to articulate, clearly, what your creative contribution is in any AI-assisted work — not defensively, but as a genuine description of where the human judgment lives
- [ ] Treat AI output as raw material, not finished work — the editing, selection, contextualization, and decision-making are where your authorship resides
- [ ] Maintain consistency between how you represent your work publicly and how it was actually made — AI-assisted is not the same as human-made, and the distinction is worth owning honestly
Tools vs. Replacement
- [ ] Use AI where it genuinely supports your workflow without replacing your creative core: archiving, drafting, transcription, administrative tasks, visual ideation
- [ ] Identify the parts of your practice where AI would reduce your work to output management — and protect those parts deliberately
- [ ] When evaluating new AI tools, ask: does this make my judgment more powerful, or does it bypass my judgment entirely?
Consent & Likeness
- [ ] Never use an AI-generated likeness, voice clone, or synthetic identity of a real artist without explicit consent — regardless of technical feasibility
- [ ] Protect your own voice, image, and creative identity by understanding what platforms you are licensing your data to and what they are permitted to do with it
- [ ] Support community norms and platform policies that treat synthetic likenesses as a consent issue, not merely a copyright issue
Memory & Documentation
- [ ] Do not outsource your cultural memory to AI systems — maintain your own archive, your own history, your own account of where your work came from and why it matters
- [ ] Use AI to support documentation — transcription, organization, search — but keep the interpretive and contextual layer human
- [ ] Invest in the oral and written transmission of scene history to younger practitioners — AI can help organize it; only humans can transmit what it meant
Community & Context
- [ ] The room context that AI cannot access — the specific community, history, and collective state of a real gathering — is your competitive advantage as a human artist. Build it deliberately
- [ ] Maintain the relational infrastructure of your scene with the same intentionality you bring to your AI tools — the human systems are harder to build and more valuable than the technical ones
- [ ] Be explicit, in your community and professional contexts, about what AI can and cannot replace — the conversation is too important to leave to people who have not spent time in actual rooms
This checklist is cultural and practical reflection — not legal, technical, financial, or career advice. The AI and creative work landscape is changing rapidly. For legal questions about AI, copyright, and artist rights, consult qualified legal counsel.
How This Opens Arc Two
Arc One built the case that culture is infrastructure: that dancefloors, artists, scenes, festivals, and DJs are load-bearing systems that carry memory, community, and possibility forward through time.
Arc Two begins by asking what happens when a new and rapidly developing technology enters that infrastructure without the cultural literacy, ethical framework, or institutional accountability to understand what it is entering.
The answer is not that AI destroys music culture. Technology has always changed music culture, and not always for the worse — electronic instruments, digital production, the internet, streaming, social platforms have all transformed the conditions of creative work, and each transformation produced losses and gains that took years to understand fully.
The answer is that the culture’s response to AI will determine whether the technology amplifies the human systems that make music matter, or gradually replaces them with plausible imitations that carry the aesthetic but not the weight.
That choice is not made by the technology. It is made by everyone inside the culture — every artist, promoter, booker, listener, platform, and institution that decides, each day, what they are willing to accept as music, as authorship, as the real thing.
The next several essays in this arc will explore dimensions of that question: what AI means for specific working contexts in electronic music culture, what human creative systems are most worth protecting, and what the arguments being made for AI adoption in music are actually asking the culture to give up.
If the argument of Arc One was that the artist is infrastructure, the opening argument of Arc Two is this: the infrastructure is now under a new kind of pressure. And the response to that pressure requires more cultural seriousness, not less.
Gig Ready is still the practical layer — the documentation, visibility, and preparation that makes individual careers resilient under changing conditions. The infrastructure argument is the same whether the pressure is gentrification, burnout, or a machine that generates plausible tracks at scale.
Get Gig Ready on Amazon → https://bit.ly/4dQLq8U
The Question Worth Sitting With
If a machine can generate music that sounds like yours — that carries the surface of your aesthetic without the years that built it, the community that shaped it, or the body that lived it — what is it that the music was actually for?
And what does the answer to that question tell you about what is worth protecting?
Richard Saxe Coulson is the author of Agents of The Universe and the founder of Nuclear Fusion Events, Enroot PR, and The Health Guru. He writes about music culture, creative technology, artist infrastructure, and the human systems that make cultural work matter.
📖 Get Gig Ready on Amazon → https://bit.ly/4dQLq8U
✉️ Subscribe on Substack → https://substack.com/@saxecoulson
🔗 Connect on LinkedIn → http://www.linkedin.com/in/saxecoulson
