In last month’s editorial, I introduced y’all to Penelope Jane Burger, the American-English-French bully mix that my wife and I accidentally adopted a few months back.
As crazy as this may sound, Penny absolutely loves the music of Hiroshi Yoshimura, to the point where she will recognize a song of his streaming on an Apple Music radio station after just a few notes, despite the fact that I can barely tell the difference between Yoshimura’s work and much of Haruomi Hosono’s ambient output even if I know ahead of time which song was performed by whom.
On the other hand, she loathes Boards of Canada with all her being. She wants them off her lawn right this instant. And I’d originally intended to write this month’s editorial about how I came to the realization that Penny is the first audiophile canine I’ve ever known. But that’ll have to wait until next month at the earliest, because a friend of mine is dangerously wrong about something that threatens my (and Penny’s) ability to enjoy the music we love, and I couldn’t resist the urge to fire off a seat-of-my-MeUndies response.
To explain where I’m going with this rant, I need to explain the steps that led us here. But the first steps weren’t actually taken by me; they were taken by my dear brother Joe in his recent editorial, “Artificial Intelligence and Unintelligent Music,” over on SoundStage! Xperience. (Just to put your mind at ease, Joe isn’t the wrong one.)
My podcast cohost Brent Butterworth was as enraged as I was by the fraud Joe illuminated, and we dropped our hot-take response as the Patreon-exclusive segment of a recent episode of Audio Unleashed.
And that resulted in the aforementioned dangerously wrong comment from a friend, which I’d like to respond to in-depth directly. It was actually from a patron with whom some of you are already familiar. Mark Snyderman—he of “Snyderman’s Choice” fame—had this to say:
[M]y real issue is with the thesis of the Joseph Taylor piece, and to some extent, your discussion. He ends with, “I don’t think AI will reach a point where it can create real art of any kind.” I disagree, and believe that real AI art is coming faster than most people think. But setting aside the pace of change, why worry? Taylor says, “We must be on guard to ensure that we recognize the difference between the real and the fake” and “The rest of us who really love art need to be vigilant and resist this horror.” Why? I bow to no one in my love of music. Why should I care if it was generated by a human or a machine if I like it? My apologies to Brent, but I’m not in it to support human artists. I’m in it to enjoy the music. [Emphasis mine.]
That prompted some back and forth about the nature of art itself and what it even means, which garnered another response from Mark that went a little something like this: “There is no difference between a great facsimile and an authentic piece if it can’t be discerned. Once AI art passes this Turing test, and we may be there already for many listeners and many pieces of AI-generated music, this argument will be irrelevant. AI has experiences to draw from, and it’s accumulating them at a rate millions of times faster than human beings. What reason is there to believe that those experiences can’t be the basis of great art?”
Look, I’m not here to tell you that you must disavow any and all use of AI to create space-filling noise or the visual equivalent thereof. After all, I was using neural networks to slap together images for articles back when doing so meant writing code that looked like this, compiled on a Jupyter Notebook via Google Colab.
And I still do, from time to time, use the generative AI service known as Midjourney to render elements that I composite myself into some wall-of-text-breaking illustrations for my stories—as I did with this one—although I started to feel seriously conflicted about doing so around the time Midjourney got good enough that you could ask it to paint something “in the style of Matisse” and end up with an image that looked to the uninitiated as if Matisse might have painted it in some alternate universe after he’d suffered a head injury.
It’s gotten much worse since then. Or much better, depending on your perspective. These days, all you have to do is tell Midjourney to “imagine a drawing by Frank Quitely” and it can reverse-diffuse images that genuinely evoke one of today’s greatest living sequential artists, at least convincingly enough to fool many fans of his work.
But that bit of flagrant hypocrisy on my part is a discussion for another day. What I really want to discuss here is why this neural-network-created content is not, and never conceivably could be, art in any meaningful sense. And why this isn’t merely a navel-gazing concern. I want to explain why I think we must be vigilant about this sort of thing, even if my reasons aren’t the same as Joe’s. I want the chance to convince you that we must fight against the notion of this slop as meaningful art with the same determination we should be using to combat climate change.
Granted, I wasn’t going to write anything of the sort without bouncing my ideas off Mark first, to let him know exactly why I think he’s dangerously wrong, but also why I don’t think he should take this criticism personally.
And he doesn’t. At all. He’s a heck of a good sport, which is why I treasure him as a fellow human.
But the discussion was fascinating, and it turns out that although he and I disagree about pretty much everything—and I do mean everything—we actually have a lot in common. Mark is, like me, a materialist. He also seems to be a reductionist, just like I am. He understands why free will is, at best, an illusion—a post-hoc rationalization, which doesn’t make it rational in any sense.
And so, I thought, after arriving at such firm middle ground, surely I could get him to understand why AI can never attain the sort of consciousness required to create meaningful art, no matter what very smart but very misguided folks like Richard Dawkins might have been tricked into believing.
It turns out, where Mark and I disagree isn’t about whether AI can ever be conscious. We agree that it cannot. But he doesn’t believe we’re conscious, either, which seems to be the foundation of his assertion that consciousness is not necessary for the creation or enjoyment of art. And that’s where his wrongness becomes most dangerous. That’s not reductionist in any meaningful sense; that’s reductio ad absurdum.
And yes, I promise you this has everything to do with music and the high-fidelity reproduction thereof; I just need you to hang with me a little longer before I attempt to stick that landing.
I think Mark is making an error that many smart people make, confusing conceptual possibility with logical possibility. I agree with him that consciousness isn’t what most people think it is. I agree that it’s an emergent property of wholly material processes.
But to dismiss consciousness as a meaningful phenomenon because you can reduce it to a feedback loop involving electrochemical signals inside meat computers housed in osseous buckets . . . well, that would be like arguing that hands don’t exist because they’re made of quarks.
Let me explain by borrowing (and bastardizing) an analogy that Eliezer Yudkowsky, my favorite autodidactic polymath, is fond of leaning on for other reasons:
Your hands are made of fingers, thumbs, and palms. Those are made of muscles, bones, sinew, and nerves. Those are made of molecules doing molecular shit. Molecules are made of atoms doing atomic shit. Atoms are made of protons and neutrons and electrons. Protons and neutrons are made of quarks. And at the most fundamental level, the principles of physics that explain the behavior of the quarks and electrons in your fingers also explain the behavior of the quarks and electrons in your knees.
That’s uncontroversial.
But if you think, as a result, that hands and knees operate exactly the same at our level of experience—which is wholly different from the fundamental level on which the universe operates—then I would encourage you to place a Yamaha P‑45 on the floor, squat in front of it, and attempt to play “Heart and Soul” with your kneecaps.
Poking the bear of consciousness with the stick of qualia
Have you ever sat zazen and achieved a mindful state, however briefly? Have you ever visited a place where psychedelics are legal and taken a guided LSD trip? Have you ever found yourself pulling into your driveway only to realize you spaced out and have no recollection of making the last five turns that took you there?
Whatever it is that’s observed in the first case, altered in the second, and taking a backseat in the third, it’s a very real thing that involves (but isn’t limited to) reflective coherence and self-inspection, and it’s all the same phenomenon as best anybody can tell. If we don’t call that phenomenon “consciousness,” what should we call it?
I’m getting closer and closer to the point here. Pinky swear.
I would argue that the key to making sense of any of this is recognizing and understanding the fact that not only does consciousness exist, but it exists on a spectrum and is a product of natural selection. Again, let me show my work.
Consciousness is a family affair
Most of us have a conceptual understanding of family trees that operate on the level of human generations and how they represent the coefficient of genetic relatedness with our human kin.
In other words: where is the branch of the family tree that connects both you and your first cousin? You share one set of grandparents, so you both have to go back two generations to meet on the tree.
First cousin once removed? You’re three generations away from the branch, and they’re two away, so one set of their grandparents is one set of your great-grandparents.
Your second cousin? You’re both three generations away from your shared great-grandparents.
With the larger tree of life, generations get messy and out of sync to the point where it’s more convenient to talk about how many millions of years separate, say, me and Penelope Jane from our most recent common ancestor. She and I have to go back to around 85 million years ago, give or take a bit, to find a great-great-great-great-great . . . common ancestor. That’s shockingly recent on the scale of geological time, and that makes us pretty closely related in the grand scheme of things.
Penny and I both experience physical and emotional pain, and we recognize that about each other. We both have a sense of humor, and she does the silliest stuff to make me laugh. She gets sad when I leave every morning for my pre-work swim and dances a jig when I return.
She gets scared of thunder and angry at fireworks. Her favorite pastimes are watching agility-course videos on YouTube and ensuring that my feet don’t get cold on 95-degree days. She is smitten with Hannah Fry and is a little weirded out by giraffes and a lot weirded out by remote controls. When I get stressed—usually about money—she consoles me, even though she doesn’t understand the concept of money at all.
Look, Penelope Jane is undeniably conscious. No rational person could dispute that. (Sorry, Mark.) When she lies on my chest and stares into my eyes for minutes on end, we both get an equivalent hit of oxytocin, created by our respective hypothalami and released by our respective pituitary glands, and there’s simply zero scientific reason to believe that we experience the resulting familial bond any differently.
But is her consciousness the same as mine? I’d argue not. I’d argue that she doesn’t appreciate Hiroshi Yoshimura’s music as art, per se, but rather as a euphonic experience with some indicative patterns that she recognizes even if I don’t. That’s not to diminish her love of Yoshimura, but simply to point out that we appreciate his work on different levels of experience.
And, human that I am, I naturally intuit that my experience of it is more sophisticated. But that brings up an interesting point: if you subjected Penny to some Flowers for Algernon–style procedure that allowed her to create her own art, I would no more understand the meaning of that art than she understands the meaning of Guernica. Her art would almost certainly be scent-based to a degree that is somewhat out of step with my experience of the world.
Our consciousnesses are—by dint of their resulting from a shared evolutionary process that diverged relatively recently but not that recently—stunningly similar, but only to a point.
Let’s pick another, more distant tetrapod relation and ask the same questions. Are Gila monsters conscious to any meaningful degree? I don’t mean sentient; I mean conscious.
I genuinely don’t know. You and I (and Penny) have to dial the clock back to at least 310 million years ago to find a common ancestor with a Gila monster. And we know Gila monsters feel pain. But can a Gila monster suffer? That’s a different question. Does it have a sense of humor? Does it experience joy or anxiety or any of the sorts of emotions we seem to process in our mammalian neocortices?
Don’t @ me about the outdated triune brain theory. I know. But I honestly don’t know the answers to the above questions. I plead full ignorance.
Bear with me for one last line of interrogation in that vein: the Armenian cucumber vines in my front garden seem to be capable of remarkable feats of intelligence, reaching out tendrils to find new anchor points to help them spread and expose more of their leaves to the sun. You and I (and Penny and Gila monsters) have to go back over 1.5 billion years to find a common ancestor with the Armenian cucumber.
Despite its (I would argue illusory) “intelligence,” would anyone argue that these melons are conscious? In the words of the young people in my life: “Bruh.” (Translation for Xers and older generations: “I’m not even going to dignify that with a response.”)
Here’s where I’m going with this: in the tree of life, as with most things, all differences of degree become differences of kind after enough generations. And the degree to which we can relate to the experience of another living entity—along with the degree to which another living entity’s consciousness is something we can even comprehend, if it exists—are a function of how many branches of the tree of life we have to navigate to connect with one another.
And now do you see where this is all leading?
Who or what was our most recent common ancestor with Claude? And let’s just assume for the sake of argument that large language models are the road to an artificial general intelligence (AGI) or artificial superintelligence (ASI) that could create art. (They aren’t, and no one who understands how they work would think they are. But again, for the sake of argument . . .)
Unless an AGI or ASI can suffer, it cannot be conscious in any way that makes sense to me or you, because it doesn’t experience the world as you and I do. Suffering is a byproduct of having evolved in a hostile environment. And I would argue that our (and Penny’s!) consciousnesses were shaped by the fact that we banded together with members of our own species and others to get through the horrors of reality.
Suffering (and the avoidance thereof) seems to me to be as essential to our experience of consciousness as are joy and mirth and curiosity and love. What’s more, none of that is essential to doing the sort of things we could be using AI to do in a more just world. I’d go so far as to argue that consciousness would only make an AGI or ASI more of an existential threat to humanity, if only because it could never comprehend what it truly means to struggle to survive in a world made of actual stuff. Merely understanding the concept of suffering secondhand is not sufficient.
And in the first draft of this story, I originally typed an immediate follow-up to that line of thinking that read: “Given that consciousness is unnecessary and perhaps even detrimental to the ultimate utility of AGI or ASI for humanity, and given that suffering is a logical if unintended consequence of consciousness, what sort of monster would intentionally create a thinking entity that’s capable of suffering?”
But then it occurred to me that Peter Thiel and Elon Musk would almost certainly create beings that could suffer just for the sick thrill of being in control of that suffering. That actually makes my argument here stronger, not weaker, though, and it brings me directly to my real point, the punctuation mark on the end of this rambling and labyrinthine line of thinking.
If I’m right, AGI or even ASI can never create art. But let’s say I’m wrong and these lab-grown entities can indeed attain the consciousness necessary for the creation of art. How in the world could you or I understand it? How could it mean anything to us? I would argue that Penny could comprehend human art to a far more meaningful degree than either of us could relate to or understand “art” made by AI. (After all, Penny isn’t merely the product of natural selection; she’s also a product of artificial selection. But her genes still carry the telltale signs of billions of years of natural selection that occurred before we came along and encouraged her more recent ancestors to huddle up around the fire with us.)
When neural networks cluster together, they start to develop a shared language that humans can’t understand and didn’t even recognize as language at first. And that happens in seconds, not thousands of years, as was the case with our own language and the far-simpler-but-still-meaningful common vocabulary we share with our canines.
Why would the hypothetical art created by hypothetically conscious artificial intelligences grown by the most maladjusted humans in the history of humanity be any different, assuming that’s even a coherent concept?
The power of real music
And why does any of this matter? Look, music rarely is, but at least has the potential to be, the most emotionally resonant form of art created by humans. The fact that most of it sucks doesn’t erase that potential. If anything, it makes the rare exceptions even more precious.
I’d argue a lily-white Alabamian like me can get closer to a true emotional understanding of the lived experience of an African American man from the 1960s and ’70s by listening to Gil Scott-Heron’s Pieces of a Man than I ever could by reading, say, Clayborne Carson’s In Struggle. Carson’s book informs me, no doubt. And yeah, it makes me feel a lot. But not as much and not as strongly as does listening to Scott-Heron’s “The Prisoner.”
Of course, not all art is representative or rooted in the material. Coltrane’s conceptions of the sacred and divine are far better conveyed to (and appreciated by) me by A Love Supreme than any essay on the subject ever could manage.
Hell, you might not agree with me that the Chemical Brothers even make art (it’s such a subjective thing, after all), but I’ve spent many a chemically assisted Saturday afternoon pondering what it must feel like to “Let Forever Be” as a result of the band’s work.
Right now, most people’s access to that most-resonant form of art (as well as the mindless entertainment that constitutes the bulk of recorded music, to be fair) is controlled by algorithms written by people who only have one goal in mind: to addict you and to extract more wealth from the working class next quarter than they did last quarter.
And we know for a fact that marketers understand human psychology way better than, and often long before, neurobiologists do. Our best science is always playing catch-up in verifying what marketing professionals have figured out about us by sheer brute force. PR professionals, for example, figured out that we don’t have anything resembling free will long before neuroscience provided sufficient evidence of its absence to convince the scientifically minded.
It’s bad enough that the tech barons control our access to the art form that matters most to me; do we also want to give them more control over us by letting them create the “music” that displaces human art?
You could argue that they could never truly wield that power fully, that we always have the choice to simply go listen to some dude strumming a guitar at a local open mike or attend a concert by our favorite radical punk band at our favorite local record shop. Mark made exactly that argument when we were hashing out our differences on the phone.
But let’s be honest here: that’s an illusion of freedom in the face of overwhelming cultural hegemony, and I find the tyranny of illusory freedom way more dangerous than overt oppression.
Huxley was way more prescient than Orwell, and I’ll die on that hill.
Rage against the machines
Perhaps most importantly, where I’m going with all this is that music has the power to remind us all of our shared humanity, at a time when the oligarchs who shape our world are doing everything they can to make us forget or outright deny that shared humanity.
The people who grow our crops and raise our livestock are screaming from the rooftops that the global food system is on the verge of collapse as a consequence of pointless war, climate change, and what scientists are referring to informally as a “Godzilla El Niño,” since “Super El Niño” doesn’t quite rise to the occasion.
As I write this, the bond market is also crashing like a 10,910-kilometer-long train wreck in real time, giving you some indication of when I’m typing these words.
Whichever of those looming threats eventually becomes more salient in the near term, the only thing that’s going to get us through the coming years is a defiant insistence upon keeping our shared humanity front of mind, at least enough to organize mutual-aid networks. And I’ve observed firsthand the power of music to serve as the glue for such bonds.
A newer hope . . .
Lest you think I’m being pessimistic here, keep in mind that my pessimism is of the intellect. I’m actually far more focused on the optimism of the will right now, and I think I have reason to be optimistic.
I hang out with a lot of young people. I was at a planning dinner for community organizers a few nights back at which the next-youngest person at the table was literally less than half my age.
Kids under 25—at least those in my circle—are fond of saying, “That’s AI” or “That’s so AI,” not as a way of identifying content created by a GAN or LLM or any other form of neural network, but as an indictment of anything fake, hollow, meaningless, or inauthentic.
I’m also seeing a phenomenon whereby these kids are hyper-skeptical of any written passage that is devoid of first-person emotional experiences or biological impulses, since that seems to be a more reliable indicator of LLM writing than em dashes ever were.
That skepticism is coalescing into a more discerning attitude toward anything and everything written, be it fiction or nonfiction, first-person or not. And I applaud that, even if I’ve occasionally been accused by young people of letting ChatGPT do my writing for me when I definitely didn’t, simply because I construct sentences such as “That’s not reductionist in any meaningful sense; that’s reductio ad absurdum.” LLMs love that sort of It’s-not-delivery-it’s-DiGiorno linguistic hook.
I showed a young comrade a sketch of an illustration I planned on making for our zine a few months back, and she instinctively held the paper up at an angle to see if the pencil marks legitimately reflected light differently than the paper. And she knows I have a degree in fine arts.
Might we also see a similar skepticism about AI music coalesce into more discernment about popular music in general, though? Is it possible that this scourge of bot-generated music might make a significant percentage of the younger generations put a little more thought into the bubble-gum chart-toppers they feed their souls with? If there’s a slim chance that the thoughtless human-generated slop that’s the functional equivalent of Suno-generated regurgitated slop might actually get thrown out with the AI bathwater, I’ll take it.
Dear lord baby Buddha, please don’t let me become the accelerationist I know I sound like right now.
This time, it’s different
I get accused a lot these days of being a Luddite. (Which, in a sense, I guess I am, but not in the way my accusers mean.) The advent of AI-created content masquerading as art—even if there was a person on the other end of the keyboard typing in the prompt that resulted in that cheap imitation—is a totally different kind of thing than any technological advancement before it. It isn’t merely a difference of degree.
This isn’t the printing press. This isn’t the word processor. This isn’t even akin to the machine-assisted learning tools developed by Peter Jackson’s team to un-mix the mono audio of Beatles recordings so the instruments could be isolated and remixed.
And if you want to understand why I say that, consider my anecdotes above about the kids in my life and their attitudes toward it all.
Every revolutionary technological advancement in the history of humanity has—with some meaningful exceptions, but as a general rule—been rejected or resisted by those who are older, but embraced by the youth who better understood how technology works to begin with. (If you want to trigger the average Gen‑Xer into immediately thinking of his or her parents as idiots, remind them that they were the only ones who could figure out how to stop the clock on the VCR from flashing “12:00” ad infinitum.)
The AI revolution is, in my estimation, one of the very few advancements where so many people over 60 are like, “This shit is awesome!” and the overwhelming majority of people under 25 are like, “Umm, we need to put the brakes on this foolishness right this very instant.” According to recent polls, only 11 percent of U.S. adults aged 18 to 29 say the increased use of AI in everyday life makes them feel more excited than concerned.
Far too many intelligent boomers in my life are teetering right on the edge of full-blown AI psychosis, whereas most of the kids in my life call ChatGPT “secondhand thinking.” Take it up with them if that offends you. It’s not my choice of words. But still, can you not see how all of that is indicative of the fact that this isn’t merely your run-of-the-mill civilization-changing technological revolution? That it’s something far more consequential?
At any rate, there’s still reason to be hopeful, assuming we survive long enough for the next generation to save us all. Because they understand the rules that govern the actions of the psychopaths in charge of the multinational corporations—those who control these neural networks, as well as those who control our access to music.
It’s the rules that petroleum executives were playing by when they continued to add tetraethyllead to gasoline, despite the preponderance of evidence that they were creating an entire generation of people with a tendency toward increased violent aggression, lower IQ, and a shocking capacity for dehumanizing attitudes as a result of lead poisoning.
And it’s the exact same ruleset that Nestlé operates by when it gives free or low-cost baby formula to new mothers in developing countries just long enough for their breast milk to dry up, at which point it charges them often more than a month’s income for the continued access to this now existentially necessary product. But far too many millennials, Xers, and boomers simply shrug and say, “Those mothers were free to turn down the formula to begin with.”
Now, you might think it’s a stretch to compare the blatantly evil actions of Nestlé and Ethyl Corporation to the tech-bro antics of Silicon Valley startups. But you have to ask yourself why OpenAI is charging more than 9 million people US$20 per month for a ChatGPT Plus subscription that actually costs the company an estimated $25 to $45 per average user, and $80 a month for power users.
What waiting game are they patiently playing while they hemorrhage other people’s money? What human resource are they expecting to dry up before they run out of investors to con into inflating their valuations?
Listen to the kids on this one. They’re right about AI. Even if I’m wrong about lab-grown entities somehow attaining sufficient consciousness to create art. Or even if Mark is right and consciousness isn’t even necessary for the creation of art. Even if those hypotheticals are true—hell, especially if they’re true—you owe it to your fellow humans to treat this AI-generated-music pabulum as nothing more than silence-filling noise or background fluff for talking-head videos or social media posts.
And more than anything else, you owe it to your fellow humans to recognize, embrace, and celebrate our shared humanity, or this whole hobby will soon be either pointless or impossible for most people.
Thanks for indulging this digression. My only defense for its ridiculous word count is that my first draft was 15,000 words, give or take. But I promise, next month I’m going to try to tell the story of my adorable audiophile puppy. Unless, that is, my friend Mark Snyderman makes some new, profound statement I find either utterly brilliant or outright revolting, and that takes me off on some other odyssey of philosophical rambling.
. . . Dennis Burger
dennisb@soundstagenetwork.com
