Where I Stand on AI and Satin Jackets

We are entering a time where AI-generated music is becoming increasingly difficult to distinguish from music made by humans. At the same time, thousands of recordings have been scraped from the Internet and used to train AI systems without the knowledge or consent of the people who created them. With a few prompts, it is now possible to generate, edit and release music without ever recording a musician, strumming a chord or even writing a single note.

I think it is worth being clear about where I stand as a music producer and composer.

All music released as Satin Jackets is made by me and the artists I collaborate with. No AI prompting is involved and no AI tools are used in the production of the music. All vocals are performed by humans and recorded specifically for the respective release. I have tested AI vocal generators, but found that if I work with a vocal for weeks in music production and listen to it every day, I want it to be human. It just doesn't feel right to me otherwise.

The same applies to the visual side of the project. All artwork for the releases is created by professional human designers. All footage used for reels, stories and videos is filmed in real life. Mixing and mastering are also carried out by human sound engineers, without the use of AI.

I find it difficult to complain about how musicians are having their work taken without consent to train AI systems, while using the same systems to replace photographers, videographers, designers or other people I work with simply because it saves money. I don't think that would be particularly consistent. And while it might save some costs in the short term, I suspect it would be fairly short-sighted in the long run.

There is one area where we have been experimenting. With the full consent of the photographers and videographers involved, we have been testing AI systems using images and video of Mr. Satin Jackets. The idea is to see whether we can find a setup that works for everyone and allows us to explore the technology without simply taking somebody's work and using it without permission. We haven't reached a final conclusion on that yet. For now, we continue to use original, human-created material.

Where I stand right now, I will not opt into licensing deals that allow my music to be used for training AI models, or to be edited, remixed or otherwise processed for AI purposes. I want the music released under the Satin Jackets name to be made by people, with people, and with the people involved knowing how and why their work is being used.

This might be perceived as conservative or even anti-technology, but that's not how I see it. For me, this is primarily about the art and the joy of creating it: the process, with all its challenges, the experience that comes from doing this for many years, and the opportunity to work with other talented people. Music should be felt by the people who create it, record it, play it and listen to it.

There is also an economic side to this that I find difficult to ignore. We are seeing a shift from paying human creators for their work towards subscription-based models that can do more and more of that work themselves, often using systems trained on the original work of the very people they are replacing. As long as I can work professionally as a freelance artist, this is not something I want to be part of. This is an individual decision and not a judgement on anyone else who is doing this differently.