Dead, Loud and Snotty

AI and the coming of slop culture.

This vulgar picture was not generated by AI. Or was it?

In 1978, when I was clinging, like a rabid monkey, to the last days of teenagerhood, I went to a club in Connecticut to see the Dead Boys, whose first album, Young, Loud and Snotty, was a favorite. Early in the show, the leopardskin spandex tights worn by the lead singer, Stiv Bators, split open, and he spent the remainder of the set with his scrotum on display. It was quite a performance, and it became one of the touchstones of my punk youth. (If there’s a pun lurking in that last sentence, please ignore it.)

The memory came back to me, unexpurgated by the years, when I read Rob Horning’s recent piece about a plan to use generative AI to re-create the voice of the long-dead Stiv on an upcoming Dead Boys album. (The band has just one original member left, guitarist Cheetah Chrome, who now also handles vocals.) The Dead Boys’ label, Cleopatra Records, announced Stiv’s second coming in a cheery press release last year:

Select tracks on the new album seamlessly blend Stiv’s distinctive voice with new elements, creating a sonic experience that feels both nostalgic and fresh. Cheetah Chrome’s vocals are artfully “dusted” with Stiv’s iconic presence, ensuring the new material retains the raw, visceral energy that defined the Dead Boys’ sound. This innovative approach celebrates the band’s history while boldly reimagining their future.

Pop goes the corpse.

In an Instagram post, Chrome explained that he agreed to the label’s plan only after undertaking a careful program of due-diligence research into artificial intelligence. He collected “info on AI to make an informed decision.” “It’s true,” Horning remarks, “nothing is more credibly punk than data-driven informed decision making.” You’ll of course recall the old Sex Pistols number, “Analytics for the U.K.”

Whether it’s the video projection of Tupac performing at Coachella or the digital staging of “ABBAtars” in ABBA Voyage or the studio fabrication of a “new” Beatles single with John Lennon singing lead, the entertainment industry has long been obsessed with bringing the dead or merely defunct back to profit-generating life. There’s nothing like a resurrection to goose back-catalog sales. As Morrissey observed in The Smiths’ “Paint a Vulgar Picture”:

At the record company meeting
On their hands, a dead star
And oh, the plans they weave
And oh, the sickening greed

The audience, for its part, is happy to overlook the ghoulishness and greed for the chance to indulge its hunger for nostalgia and spectacle.

Using AI to generate the voice of a dead singer is at one level just the latest twist on this market-tested scheme. It’s a stunt. It’s hokum. But it portends something bigger. With the production of so-called “AI slop” — mediocre AI-generated songs, videos, images, and texts — becoming ever easier, the resulting content is flooding online media platforms. Much of the slop is at the moment getting attention mainly for its novelty. Shrimp Jesus! Haha. But as the content takes on a more professional sheen and gets algorithmically tailored to individual preferences, it could easily become a major component of media feeds. Maybe the major component. Already, according to one estimate, half of the longer posts on LinkedIn are penned by AI.

Horning reports on a recent festival of AI-generated videos hosted by the Chinese TV manufacturer TCL. The company foresees such videos, all pretty crappy, forming the core content for its strategy of providing owners of its sets with “a lean-back binge-watching experience,” according to one of its executives. “Data told us that our users don’t want to work that hard. Half of them don’t even change the channel.” Because the videos feel like the digital content people are habituated to seeing online, Horning suggests, drawing on Jerry Mander’s earlier writings on TV programming, they fulfill viewers’ expectations for diverting entertainment. They may, to quote Mander, “contain no life at all,” but that doesn’t matter as long as they offer a reasonable simulation of what the consumer is used to consuming. Slop’s good enough.

And slop is cheap. Discussions of feed algorithms tend to focus on the demand side — the matching of a bit of content to an individual consumer through the instantaneous analysis of the triggers of that consumer’s behavior. Less attention has been given to the supply side — the sourcing of bits of content through the instantaneous analysis of the cost of the content to the platform. In contrast to traditional media companies, which produce and distribute a fairly limited set of offerings, social-media platforms operate vast, complex cultural supply chains that have to be optimized to generate profit through a multitude of tiny informational transactions. Because they deliver billions of bits of content every moment around the clock, it’s imperative they find the cheapest possible content to feed to consumers.

Up to now, as I discussed in my post “Dead Labor, Dead Speech,” most major platforms have exploited freely contributed content from users — posts, comments, photos, etc. — to fill their feeds at little or no cost. But even platforms that don’t have access to user-generated content are relentless about finding cheap sources of supply. Few people were surprised by the recent report in Harper’s that Spotify pads popular playlists with entirely fabricated “ghost artists” whose songs are mass produced in the musical equivalent of content farms. Because the songs, again, feel like the genres listeners are familiar with, they stream by unnoticed. Cue up Chill World Vibes or Lazy Sunday Morning, and let it flow.

What Spotify has been engaged in — and it’s hardly alone — is a large-scale experiment to test the fungibility of culture. How far can we go in replacing creative work (and the artists who create it) with manufactured slop? With generative AI, the scale of that experiment is going to get much, much larger. By automating content farming, platforms will be able to further drive down the cost of content — and further reduce their reliance on actual artists. More than that, they’ll be able to generate the content in real time, custom-fitted to individual demand. The supply is unlimited.

AI slop is a sort of liquid nothingness that will pour through culture, searching for whatever’s fungible and displacing it. What’s really being tested here is human taste. Will we accept a simulacrum of a work of art or craft as a satisfactory substitute for the real thing? Will we even notice the difference?

Morrissey is a better preacher than I am, so I’ll hand the mic back to him:

But you could have said no
If you'd wanted to
You could have said no
If you'd wanted to