A couple of weeks ago I participated in a Zoom panel discussion about the use of AI (artificial intelligence) software in the arts. This was hosted by Nick Batzdorf, who is the content director at the Synth & Software website, and it’s now available (or at least the audio is) on that site.
It was an interesting discussion, and a number of points were touched on. I got a little irate about the need to know music theory. I don’t like it when I get irate; I sound like an old guy shouting, “You kids get off of my lawn!” What upset me, I think, was that while Olivio Sarikas, who had a lot to say during the podcast, is a fan of AI, notably of an image-generating service called Midjourney, he doesn’t seem to know much about music.
I have no firm opinion about Midjourney except from a philosophical perspective, but it’s clear to me that the process of generating appealing images using software tools is different in some basic ways from the process of generating appealing music using software tools. From the word ‘go,’ images have content. Music data is entirely abstract. And because the millions of starter images uploaded to Midjourney already have content that the human eye and brain can interpret in meaningful ways, the machine doesn’t have to do as much to fit things together. To be specific, the machine doesn’t have to know what the meaningful content actually is. Since music is an abstract language, it requires a higher degree of human perceptiveness and intervention.
I’m pretty sure a human who is using Midjourney is also engaged in finding the most meaningful products that the AI has generated — picking one image that seems especially apropos and tossing out five dozen others that don’t quite do the job. Or at least I hope so. But I do maintain that an AI that combines images in various ways is a lot more likely to come up with something that is meaningful in human terms than an AI that is tasked with combining musical ideas.
But that’s not what I want to talk about.
Early in our discussion, Olivio was talking about how there’s a greater need today for images that can be created quickly. He referred to the use of cameras (you remember cameras, I’ll bet). Once upon a time, you had to take the film into a darkroom, develop the film, make contact sheets, choose the images you wanted to enlarge, go back into the darkroom to print the enlargements, then half-tone them and send the half-tones off to the printer. This could take a week. In the internet age, people need to be able to create and upload an image in half an hour.
That’s true, but he framed it, quite correctly, in economic terms. Corporations want to reduce costs. If a process is labor-intensive, it’s also expensive. If a software tool can produce striking imagery in half an hour, and if no one with an expert eye has to be employed, there’s a significant cost savings.
But is the result art? Well, no, it’s not.
Art is the process that the artist goes through, mentally and emotionally, in creating the work. I’ll repeat that: Art is the process. It’s not the outcome or the product. Art is not and can never be a product. If you have a brilliant piece of software that can produce eye-catching imagery at the push of a button, what it’s producing is not art; it’s processed art substance. It’s a fraud.
I don’t like the word “spiritual.” I try never to use it. But I think it’s good shorthand in this discussion, so I’m going to set aside my reservations for a minute or two. When the person tasked with producing the art, be it image, music, or what-have-you, is not intimately engaged from moment to moment in every decision about what is to go into the work of art, the person at the helm of this metaphorical boat is not an artist at all. The person at the helm is spiritually bereft.
And what’s worse, anybody who then encounters the work of “art,” no matter how compelling it might appear to be, is also spiritually deprived. Cheated. Denied the full experience of their humanity.
I once asked Wendy Carlos if modern music technology had made it easier to compose music. She said no. She maintained, I think correctly, that the process of creating new music is as difficult now as it was before. So I think Wendy would agree with the point I’m making. The process of creation is an interior process in the artist, and requires both thought and feeling. You can’t offload it to a machine.
Also, it takes time. If you produce art too quickly, either because you’re under financial pressure or because you’re just too lazy to do it properly, you’re not an artist at all. Maybe a bullshit artist, but nothing beyond that.
Also, there’s no excuse for not knowing the theory, the aesthetic underpinnings of your medium, whatever they happen to be. That’s not quite the same thing as saying all composers of music need to know the conventional harmony theory that was developed in Europe between the 16th and 19th centuries. There are many theories about music, and Western harmony is only one of them. But if you’re planning to write music that uses chords and melodies within the European tradition and you think you don’t need to know harmony theory — or worse, if you imagine that harmony theory will somehow inhibit your awesome creativity — you’re just being stupid. There’s no reason to sugar-coat that. I said “stupid,” and I meant it.
But that’s just “you kids get off of my lawn.” My central point is that art is not a commodity. It’s something that happens within the artist. You can’t offload it to a machine.
We have to acknowledge that, unlike the visual and literary artist, music has always been a collaborative art. Beethoven did not play all of the instruments in the orchestra, nor did he build the instruments! When I compose music in the computer, I’m quite willing to collaborate with software developers and sound designers I’ve never met, whose names I may not know. And yes, I sometimes use loops. But the aesthetic decisions are always mine, and they may take time. A loop may need to be edited in some way in order to be usable. I need to know how the editing tools work, and I also need to have some inarticulate feeling about what’s working and what’s not.
If you’re using a drum loop as a short cut because you don’t know how to make good beats, you’re cheating. After finding a good loop, you may need to work with it in arbitrarily complex ways in order to make it fit within your own artistic vision. Filtering. Trimming. Quantizing. There are dozens of ways to work with a loop, and if you’re composing into a computer you really do need to know them all. Or the loop may be perfect as is, but you’ll still need to surround it with other carefully chosen components.
Listening closely and learning how those components fit together takes time — and not just a few hours or a few days. It takes a lifetime.
Any of your musical experiences, as either a listener or a player, may become relevant as you listen to your mix for the 20th time. If your bank of past musical experiences is slim, your music will be shallow. If you imagine you can slap together some loops and then upload the finished track in an hour or two, I don’t even want to talk to you. You’re not a musician. Go away. Or at the very least, get off of my lawn.