Is AI All Bad in Music?

June 2026. How I use and not use AI in creating music videos.

The image above is created by use of Midjourney AI.

AI-generated productions are currently (S26) facing resistance from many of the people who consume creative works. Music and images created by—or with—the help of artificial intelligence are often looked down upon.

I can understand the logic if, as a consumer, you value the artist more than the artwork itself. It's difficult to become a fan of a cold AI model. If you care more about the finished product, however, things become a little less clear-cut. AI produces a lot of junk—especially in music—but it can also create something worth watching or listening to. Exactly like human artists can.

Most AI systems have been trained on vast amounts of music and images they were never given permission to use. That's a legal issue. Perhaps a moral one as well. Are you dealing in stolen goods if you use works that were effectively taken from others as the foundation for your own creations?

That said, we shouldn't pretend that human artists aren't also trained—or "inspired"—by the work of others. Pablo Picasso famously said, "Good artists copy, great artists steal." In most cases, however, people have paid in some way for the experiences from which they draw inspiration.

IMAGES AND VIDEO CLIPS ARE HARD WORK

When I use AI for my music videos, it's a long and demanding process. I usually have a clear plan beforehand regarding what I need and how I want it to look. When you feed those requirements into a prompt (the text instruction you give the machine), you rarely get what you need on the first attempt.

The machine may produce something beautiful or amusing, but at best it can often be saved as an idea for another project. More commonly, the results are mediocre or far from what I had in mind. You end up modifying and expanding the prompt over and over until you get closer to the desired result. Often, you have to switch to a different AI model entirely.

Almost every time, I still need to finish the images in Photoshop. That's actually fine by me because, formally speaking, that makes the image my own work. In much the same way, incorporating it into a music video transforms it into a new work over which I hold the copyright.

Why spend so many hours working with a difficult collaborator like generative AI? Because the alternative would often require a large team, professional models, expensive equipment, and rented locations—resources I simply don't have access to.

LANGUAGE MODELS AND LYRICS

My lyrics usually emerge from some strange corner of my subconscious. Until recently, I refined them using online synonym and rhyming dictionaries, which thankfully can also count syllables.

Lately, I've been using ChatGPT as a translator and rhyme assistant. The finished product remains entirely under my control. ChatGPT is a chatterbox, but it does know a thing or two about the English language.

THE MUSIC IS AI-FREE (IN THE USUAL SENSE)

I compose music from my own imagination. If I steal anything, it's certainly not intentional. Working on music makes me happy—that's the driving force behind it. Creative therapy, if you will. For that reason, I have no intention of using AI to compose music.

I make that qualification because the software I use includes software agents and virtual session musicians that can perform using built-in instruments—parts that I could, in principle, play myself.

That's very different from what happens on platforms like SUNO, where the entire soundscape is generated by AI in one go. In fact, it's much closer to what has long been standard practice in pop and rock production, where highly skilled session musicians often perform parts that might otherwise have been played by members of the band.

Take drums, for example. I don't own a drum kit, and I'm not particularly good at maintaining a steady rhythm. However, Apple Logic Pro contains an extremely intelligent virtual drummer. It can do a great deal on its own. But I decide the drum kit's configuration and brand, how it's tuned, where fills are allowed, how much swing the groove should have, and countless other details—exactly the sort of instructions a producer would give to a real drummer.

On top of that, I use a similar drum kit that I "play" myself via a MIDI keyboard or by directly editing the notes whenever the virtual session drummer doesn't have a clue.

With SUNO and similar services, the music comes out fully recorded and mixed, ready to use. Personally, I don't find that very interesting. What fascinates me is shaping raw recordings into a cohesive whole and polishing them so they sound good on hi-fi systems, car stereos, and headphones alike.

There are AI tools for this too, and they are widely used—especially for the final stage known as mastering. But they have no place in my workflow. They tend to make everything sound the same.

AI IN THE CREATIVE PROCESS

So yes, artificial intelligence is part of my creative process whenever it makes sense for me to use it. Others are free to make their own choices.

If you dislike AI on principle, or because of a perfectly legitimate concern or fear, that's entirely your right. But I wish you the best of luck in trying to avoid it altogether.

The manuscript for this article was converted to English by chatFPT and then edited afterwards by a human.