AI Artomat

 

I don’t like the idea that creativity has to be convenient—they are two terms that don’t go together. We want things in a creative environment to perhaps be a little bit easier to do, but that’s what tools do: If you have a tool that is well-designed and doesn’t hurt your hand when you use it then that’s an element of design. In many ways, music has to be “designed”.

The more we do something the more we can get out of it but the question is whether the things that we’re doing with a certain technology by forcing a result, results in something that’s actually good. This is something that I’ve been observing with artificial intelligence in general over the past decade. I was initially interested in it as an extension of generative music, but the more that they hype it, write articles about it and share it on social media it looks like it has legs—that this is going to be a promising technology.        

The articles that I’ve been reading lately are covering kind of the same territory as they have over the past five years. There’s been really no development at all. I do like the Boomy app—which is essentially a randomness generator on sample libraries organized into categories, which you can move around on a (crude) timeline. But I see it as raw material, not a finalized piece of music. I like to play guitar, bass, or keyboard against it and attempt to shape it into a viable musical idea. It might work out of the box, but usually it doesn’t. No one would be satisfied with the result.    

One of the complaints I have about AI is that it promotes the idea of convenience in creativity. But what else are you going to do with your time? If it takes five minutes to write (generate) a song and release it on Spotify how satisfying is that? You want to have something that you can spend three to five hours on—or a week or a month. #riff 

9/6/2021    

[9/6/2025: From a recent article about the new AI pop star imoliver (Oliver McCann): “Once Smith gets inspiration, it takes him just 10 minutes to write the lyrics. But then he'll spend as much as eight to nine hours generating different versions until the song "matches my vision." McCann said he'll often create up to 100 different versions of a song by prompting and re-prompting the AI system before he's satisfied.” It certainly can take a long time to “curate” a song or album, but it’s just that–curation, not composition. It’s like generating an art exhibition where everything is ordered on Amazon, art, framing, and shipped to the gallery and hung for an extra charge. All you have to do is generate an artist statement and show up for the accolades. It’s the 21st century version of Warhol, but Warhol physically worked very hard making the silkscreens. But the 60s also included artists creating just the concepts and not making anything–a very Duchamp idea, so it’s a continuation of that. The difference is an awareness of your intentions, as opposed to appearing that you really want to be a celebrity, and the convenience of AI allows you to do that. Conceptual artists in art history were typically crazy like a fox, which imoliver doesn’t appear to be. Real artists can pivot to new periods. If all you produce is generative art, how do you follow that? I would assert that it’s not art writ large unless there’s some aspect of craft or philosophy–at least in retrospect. As to lyric writing, it can take me a long time because I want them to reach a certain level where they work with the music in the best way possible. It’s like making a painting where each mark or gesture makes a difference–up until the point you know you’ve finished it. AI generation can be like that, so it’s similar in that regard. But again, where is there to go for the next period? It won't be neoclassical because you were never “classical”. Or it could simply be you don’t care and you’re ok with being flash in the pan. But I see that as particularly reprehensible].