I Spoke Without Pay at an AI Conference Charging $375 a Head and No I’m Not Bitter At All

Tod Machover delivering a keynote address at the AES/Berklee Symposium on AI and the Musician

AI doesn’t do irony, I’ve heard. Last week, at a symposium on “AI and the Musician” sponsored by the Audio Engineering Society and Berklee College of Music, keynote speaker Tod Machover said that Generative AI is also not good at musical structure, emotion, performance and interpretation, adaptability, or cultural sensitivity.

Other than that, Mrs. Lincoln, how was the show?

The irony of this AI conference – the part that AI would be incapable of perceiving - was that every statement boosting the technology could also be taken as a critique. Composer and MIT professor Tod Machover, in his keynote, played examples of cutting edge Generative AI music that he pronounced, “pretty [pretty] good.” But why we need a new technology to produce mediocrity, he didn’t explain.

I admit I lost patience quickly with the tech idealism that permeated the place, and after a couple more presentations of pretty good achievements by AI, I headed out the door to wait elsewhere for my own panel. It was a beautiful day, so I went to nearby Fenway Park and took a tour of the field with a group of sunburned tourists. Afterwards, I stopped in a convenience store near the conference center and bought Advil because I had developed a splitting headache.

The convenience store was in a space that used to be one of my favorite local guitar shops. (This is another example of irony, in case you are AI.)

View of Fenway Park from top of the Green Monster

I ran into music tech scholar Mark Cartwright, who had invited me to join the panel he was moderating. “I have a headache,” he said. I offered him the other half of my packet of Advil. He swallowed it and we went into the meeting room which was darkened for slides, quiet and cool but airless in a recently renovated way. Mark had assembled an interesting group for our panel: a historian of early automatic music machines, a researcher on the voice with an appointment at Mount Sinai hospital, a Grammy-winning hip hop producer, a tech designer for Native Instruments, and me. (Punk rock drummer, kind of? In this context, that’s how I felt at least.)

Our panel proceeded to introduce a number of cautionary notes to the general air of tech idealism – the historian reminded us that there’s nothing new about the pursuit of automation in music; the researcher on the voice made powerful statements about using tech to isolate aspects of music from the communities that gave them meaning; the hip-hop producer testified to the ever-shifting technologies of music and how creative individuals will invariably put them to unforeseen purposes; and the engineer for Native Instruments, god love him, said our panel was opening his mind to complications that hadn’t occurred to him in his daily work of solving tech problems.

My own contribution to our talk about “AI and the Musician” centered around this idea:

Have we already created the conditions where only AI can be a musician?

It’s not a new technology making this the case in our society. It’s existing and increasing consolidation of capital, and denigration of labor, that makes it so hard to survive as a working musician in this moment. And this is the moment into which AI is being introduced as a useful technology.

Whatever AI is capable of in music – even if it never gets more than pretty good – it is supremely well suited to making music in a culture that does not reward labor. Content machines are a dream of corporate platforms, and corporate platforms are a dream of venture capitalists. AI is the answer to a problem of scale encountered by a platform economy unwilling to center itself on labor.

As our group left the dais, I heard the moderator of the conference transition to the next panel: “Well, now that we’re all thoroughly depressed…” he said. Ironically.

Listening to: MESTIZX by Ibelisse Guardia Ferragutti & Frank Rosaly

Cooking: Rhubarb-orange jam