I am not one hundred percent sure where I sit on the AI music debate. What follows is probably going to be a bit of a hot take. I am not even entirely sure if this is my permanent stance on the matter. But for those of us who like to think a little deeper about this stuff, consider this a thought experiment.
If you spend any time in music production circles online, you will see a massive, existential panic surrounding platforms like Suno and Udio. The prevailing argument is that AI-generated music is inherently soulless because the root cause of the concern is a lack of humanity. A machine made it, therefore it lacks the human element.
But as an electronic music producer—and specifically, someone who leans heavily into hardware sequencers and modular synthesis—I find that “human element” argument a bit hypocritical.
We have been using machines to make musical decisions for us for decades.
If I set my Oxi One sequencer to a Minor Pentatonic scale, turn up the probability engine, and literally just hit a “randomize” button until a cool bassline pops out, how is that fundamentally different from typing a text prompt into an AI generator? In both scenarios, the machine generated the notes. I simply curated the result.
So, where is the actual line between generative music and AI? And why does one feel like a valid studio technique, while the other feels like an existential threat?
The Training Data Divide
The most obvious, objective difference between the two is the ethical reality of how the machines learn.
When I patch a Turing Machine module in my Eurorack rig, or use a Euclidean rhythm generator, the machine isn’t referencing other people’s music. It is operating on pure, localized math. It is executing a probability algorithm based on the voltage and the clock parameters I have physically fed into it.
Artificial Intelligence, on the other hand, is built on the scraped, digested, and assimilated work of millions of human artists. It doesn’t understand music theory or voltage; it understands patterns in its training data.
I deeply appreciate this distinction. Generative music is built on math; AI music is built on the uncredited work of others. For many people, that ethical breach is where the conversation permanently ends.
But if we set the copyright argument aside for a moment and look purely at the process of creation, the line gets incredibly blurry.
The Proxy of Effort
This brings us to a much more uncomfortable philosophical question: Is one method actually “harder” than the other? And is “how hard something is to make” a worthy proxy for how good it actually is?
We have a deeply ingrained bias that hard work equals artistic value.
I wrote recently about the “intentional inefficiency” of my hybrid studio. I like the fact that patching a modular synthesiser takes ten times as long as loading a VST. I enjoy the friction, the troubleshooting, and the physical resistance of the cables.
But if I am being totally blunt, a listener on a dancefloor does not care how hard I worked. They are just there for the vibes. They do not care if it took me three weeks to meticulously program a hi-hat pattern, or if I generated it in three seconds by hitting a random button. They only care if the track makes them move.
I see this exact same frustration boiling over in the DJ community right now. You go online and watch endless complaints about modern DJs being in it for the ego, or the fame, or the ridiculous fees. But doesn’t all of that anger stem from the exact same uncomfortable truth? The consumer doesn’t care.
We, as producers and DJs who are deeply passionate about the craft, are desperately projecting our own high, purist standards onto an audience that just wants to dance. We want them to appreciate the effort. But they don’t.
And honestly, does it really matter if it doesn’t matter to the people we are trying to present to? Why do we even care what they care?
If a producer spends three hours writing, refining, and tweaking a highly specific, complex text prompt to coax a beautiful piece of music out of an AI, why is that considered “lazy,” while me hitting a randomizer button on a sequencer is considered “art”?
Perhaps our discomfort with AI isn’t actually about the lack of humanity. Perhaps we are just terrified that AI exposes the fact that effort doesn’t necessarily equal quality. It strips away the romanticism of the “struggling artist” and delivers the end product with zero friction.
The Architect vs. The Consumer
So, if effort isn’t a reliable metric for artistic value, and if both methods rely on machines to generate the notes, why do I still prefer the hardware randomizer over the text prompt?
I think it comes down to the difference between building a system and ordering a product.
When I use generative tools in the studio, I am the architect. I decide the tempo, the scale, the sound design, the routing, and the parameters of the chaos. I build the fences, and then I let the generative algorithm run wild inside that paddock. I am collaborating with the machine in real-time, reacting to the voltages it spits out and physically turning knobs to shape the outcome.
This real-time collaboration is exactly why generative hardware is far superior at creating “happy accidents.” When I let a sequence run wild, I frequently find myself in a musical place I never expected to be. The machine throws me a curveball, and with a bit of physical coaxing and knob-turning in the direction I want to go, I arrive at a result I never would have come up with on my own.
AI, on the other hand, feels stubbornly subservient. When you use an AI generator, you are a consumer. You place an order at a digital drive-through. You give it a prompt, and it tries its absolute best to create exactly what you asked for. The element of surprise is missing because the AI is attempting to perfectly fulfill a precise order, rather than suggesting an alternative path. (Though, if I am being totally honest, maybe my prompting skills just need some work).
The Integration and the Original Sin
Of course, we are already seeing the lines blur even further.
Platforms like Suno are no longer just simple text boxes where you type a genre and wait; they are pushing deep into the studio space, even offering their own DAW-like environments. You can get incredibly granular and in-depth with how you prompt, arrange, and manipulate the AI generation.
But even with that added layer of control, it still feels like there is an intangible difference.
It raises a massive question for the next few years: As AI becomes more seamlessly integrated into the physical process of making music—when it inevitably just sits inside Ableton as a native tool—will it become broadly accepted? Will we eventually view it the same way we view a complex generative sequencer today?
Or do the ethical and moral issues of the training data trump everything else? For many musicians, it doesn’t matter how useful or deeply integrated the tool becomes. The original sin of how the machine learned to sing will permanently prevent them from ever wanting to use it.
I don’t know if AI is inherently evil, or if it will destroy the music industry. But this entire debate has made one thing abundantly clear to me: I should be doing this for the love of the craft, not for the validation of the consumer.