Why AI Will Kill Artistic Creativity

The coming age without relevant human art

When I began this substack last year, my intention was to write mainly about the arts: Music, literature, painting, etc. I knew that there would necessarily have to be overlap with cultural and societal issues, though I tried to limit these, partly so as to dwell more on the joyful aspects of art that I enjoy. Perhaps the election results are mainly to blame — my most recent essays have been mostly of a political nature — but I find myself in a gloomy place of considering what’s to come next for humans. So, today’s post is a sort of synthesis of these two broad topics. I think most of us have vastly underestimated the changes we are going to see — perhaps very soon — as artificial intelligence (AI) develops. The essay below explores how I think AI is likely to impact the future of human art. Much of what I write may be considered controversial, and I invite all your comments and thoughts.

Warning: Strong opinions ahead. Read at your peril.

I recently found a way to improve my sleep, and therefore rather dramatically to improve my overall well-being. There’s no reason to go into details, except to say that for most of my adult life, I have suffered from poor restorative sleep, in part due to severe sleep apnea. Since landing on a fix that seems pretty effective so far, I have become aware of just how adversely my poor sleep was affecting my entire life. My point here is that when circumstances change slowly, such as my deteriorating sleep in my twenties and thirties, it’s very difficult for us humans to be aware of the change. The frog in the slowly heating pot of water. Humans are famously adaptable, and I found ways to cope with my lack of sleep, though I now believe that certain things might have gone better for me had I acted to improve the situation much earlier. Only now, with a quick improvement in the situation, can I understand to what extent my life was affected. This is how things go for us, both as individuals and, I propose, as societies.

Similarly gradual changes can also happen on the cultural and societal levels (though they can also happen suddenly). This essay is a warning to raise awareness — so maybe we can avoid the frog’s fate — about what’s in store for us as artificial intelligence takes over the portion of our culture we know as the arts (which this Substack is mainly about) and to warn against the kind of complacency I exhibited regarding my poor sleep. It seems, therefore, sensible to take a look at where we stand at this moment before AI has insinuated itself broadly, and how we got here.

The recent bunch of novels I’ve read have all been from the second half of the eighteenth century, which I regard as the golden age of the novel. Melville, Hawthorne, James, Twain, Eliot, Dickens, the Brontes, Hugo, Conrad, Hardy, Balzac, Dostoyevsky, Tolstoy, among many others. They all wrote from about 1850 to the early twentieth century. (I emphasize at the outset that there are, of course, no hard cutoffs; rather, I am remarking on a general period where the novel flourished. Poetry, too, for that matter.) Identifying this time of flourishing begs the question: Why that period?

There must be a multitude of answers to that question, but I’ll offer up just two here: First, every art form has an arc to it, and thus a rise as it develops out of the dust. In the development of music, for example, JS Bach, then Mozart, then Beethoven, did not spring from nowhere. A hundred years before Bach and his codification of four-part harmony, plainchant was still pervasive. Harmony was developed gradually, in fits and starts. Linear perspective had to be discovered (invented?) by Filippo Brunelleschi in the fourteenth century to usher in the great developments of painting in the sixteenth and seventeenth centuries. It takes time to develop the techniques that in turn serve expression. In my own songwriting experience, I have had to develop certain specific technical skills on guitar in order to play what I want to express. It’s a big part of how art works. And a great thing about this is that once a technique is out there, it’s out forever. The world cannot unsee perspective in painting, nor unhear harmony in music. Sometimes, such innovation takes time to be accepted broadly -- and sometimes, as for example, atonal music, it never is -- but the key here is that artistic innovation takes time, and development of the novel form matured and rose, in my opinion, to its zenith in the second half of the nineteenth century.

The second reason I propose for why the novel (simply the example of an art form I’ve selected to illustrate a golden period) rose to the highest level then is cultural. By this, I mean that — given that sufficient techniques were developed — cultural conditions were in place for this art form to be received by significant numbers of people. The most important of these conditions with respect to the novel was, I think, burgeoning literacy among the populace. Other conditions were also in place: More leisure time among the middle classes, cheaper printing costs making books and magazines more affordable, and most notably, the absence of many other modes of entertainment, especially the surfeit we have today. Of course, there were all sorts of recreations available in the nineteenth century, but reading -- including, importantly, reading aloud to one another in the evening (cf. Little Women, another great novel from the period we’re discussing) -- was among the most broadly practiced. There was a hunger for it. And that hunger -- that market -- naturally gave rise to a large number of writers to address it and satisfy that hunger.

This last point now begs the question: What caused the decline in the novel? (At this point, perhaps many of you will read the last sentence with skepticism about the novel’s decline. I had dinner last week with someone who insisted that Barbara Kingsolver will be regarded a hundred years from now among the greatest of novelists, to be included in the list above. Perhaps so — I have not read her, so I can’t comment in any specific way. But to my way of thinking, several generations — three or four, at least -- need to pass before anything is declared to be an all-time great. Maybe Kingsolver will be considered that way, but on general principle, I’d say that the odds are enormously against it.) One answer to why the novel has declined in the artistic sense is simply money: Publishers are motivated to make money, and so they invest in formulaic works that have proven to be good sellers. Think of Steele, Robbins, King, Clancy, Koontz, etc. My guess is that even these writers don’t consider their novels on the same level as Tolstoy’s (William Faulkner, when asked what he considered to be the three greatest novels ever written, replied, “Anna Karenina, Anna Karenina, and Anna Karenina.”)

My second answer to this question is what Ira Gershwin playfully wrote in one of the all-time great songs (“Our Love Is Here To Stay”): “The radio, and the telephone, and the movies that we know. / May just be passing fancies and in time may go.” Replace telephone in that list with television, and the demise of the novel can be seen to have been inevitable. These forms of entertainment crowded out space for reading, and therefore, for novels. They also — along, famously, with today’s social media — eroded our attention span, to the point where practically no one reads anymore (among younger people, almost literally no one). The novels that remain coming out are, like today’s music, extremely formulaic, designed to excite pleasure centers in the brain, rather than be original works of art that satisfy the artist’s need to write. In this environment, the exceptional writer (or songwriter) can almost never break through; Once again, publishers generally want to sell books, so they are selective on the basis of what they believe will sell. This is not art; it’s factory farming.

As I just hinted at, today’s music industry is another great example of the devolution of the once great art of songwriting into an “industrialized process.” In his book, The Song Machine, John Seabrook explains the process by which a handful of Swedish songwriters have been cranking out pop music hits for over twenty years:

They do it in a way that I call “track and hook.” … The usual method of writing songs is that it starts with a melody and a lyric writer, then the production gets added later, whereas in this method, the production is first. The producer makes a track on the computer, which is a chord progression, beats, and then some instrumentation. Then that track gets sent out to a bunch of different “top line writers” or “hook writers,” and sometimes you’ll send the same track to a bunch of different melody writers, and they’ll listen to it and then add hooks, which are short melodies. They’re not flowing, long melodies like a Burt Bacharach melody, but a very condensed melody, like a phrase. And then the producer will listen to those, and if he likes one, finish producing the song, and then the lyrics will come at the end. The reason that that method has been so successful is basically you can produce a lot more songs in a shorter amount of time.

You may disagree, but I cannot consider this art. More factory farming. Now, you may say, “well, there are still ‘real’ songwriters who work by a less formulaic process,” and you’re right. I’m one of them! But in the stream that Spotify’s algorithm sends out, these rare songs are entirely overwhelmed by the factory farm products, just as factory farm food overwhelms more wholesome alternatives. Especially when, like ultra-processed food, they’re designed to appeal to the most easily excitable senses, what’s now being called “dopamine songs.” What chance does real art have? Practically none.

The music case illustrates the “light,” purely human version of what’s coming from AI. AI will be factory farming of art on the greatest scale, and will overwhelm everything else. What makes me so sure? For one thing, the formulaic process (NOT art) has already been proven to work, in the commercial sense. Resoundingly (from a business standpoint, which is really the only remaining driver). Even more perniciously, though, AI lives in a hall of mirrors that can only and must flatten out all its output. AI knows only what it consumes and creates, so its “creative” output is necessarily highly derivative and “samey,” which matches what I’ve seen and heard so far (discounting its “hallucinations”). The pop-music discussed above is also highly derivative and samey (for years, my sense has been that it’s been steadily moving in those directions), and as I mentioned, this is only the light version of what’s to come with AI songs and other “creative” output. The hall of mirrors effect will accelerate this to the point where there will be little if any originality. Already, human creative output has largely lost originality: Just note that almost all movies made these days — the ones that make any money, anyway — are either franchises (e.g., the Marvel, DC, and Star Wars universes), sequels, or remakes. It seems that movie-makers have been stripped of imagination and originality, though I think it’s more accurate to say that they are responding economically (movie studios have been taken over by private equity and multinational corporations whose only motivation is the bottom line) to an audience whose cultural taste has been smashed to nihilistic dreck, and which continues to spiral downward.

Originality is the life-blood of art. A better metaphor is that it’s the random mutation in the DNA of art that keeps it adaptable to environmental changes. Just as most DNA mutations are more harmful (or neutral) than helpful, most originality in art has always been rubbish. That’s not only ok, though -- it’s essential. When the first flat seventh was sung by slave fieldworkers in Louisiana, it would have sounded like dissonant rubbish to “more refined” ears, not as the mournful cry it was. This was a mutation that survived in an enormously consequential way. I won’t bother to go through a list of analogous innovations — mutations — in every art form, as I think it’s clear. It’s part of why we talk about genres. AI, living as it does in its increasingly stagnating hall of mirrors is like inbreeding on a massive scale, which means that art — AI-generated “art,” at any rate — will not survive, just as in-bred species don’t. Humans will continue to create — it’s a central part of what we do — but it will be so crowded out by the degenerate slop produced by AI that it will be a long time, I believe, if ever, before real human art again becomes relevant.

Human brains work entirely differently than will any in silico consciousness. Among other factors, in silico consciousness will not respond to its “body” in any way similar to us, nor is it likely to respond as we do to personal matters or cultural events that drive so much human art. It will be a very long time (my bet is on never) before we figure out how to design in silico consciousness that is closely aligned with our own. Until that happens, I can’t see that any creative output can be at all similar between the two.

What I think we need to reckon with is the looming losses that we will be experiencing as AI “art” takes over from and crowds out human art. The scale, breadth, and depth of human artifice for at least the past fifty thousand years is a central part of what defines us. It surely has been one of the most important things that still binds us to the natural world. I fear and shudder for the humans of future generations who will have a diminished or perhaps non-existent capacity for producing art as we have known it. Perhaps the fight for significant regulation on AI could mitigate these losses, but our current political climate makes that appear to be extremely unlikely. You may say I’m being hyperbolic, but sadly, I see the death of human art coming fast.