DJ X: The Perils of Personalization in a Post-Radio Era

Design by Etai Smotrich-Barr

When Spotify rolls out a new feature, I am the type of person who is dying to get their hands on it. This week’s newest feature is buried––one can only access it on the mobile app, hidden under the music only subheading. It’s mysterious too, simply called DJ, and presented as a blue box with a neon green circle inside. 

This new feature, it turns out, is an Artificially Intelligent pocket-sized disc jockey. Meet Your DJ, Spotify advertises. 

Trained on an algorithmic understanding of a user’s individual taste, as well as a generative use of OpenAI, Spotify’s DJ intends to expand the possibilities of a personalized listening experience. “Personalization is at the heart of what we do at Spotify,” the service explained. “We’re always looking for innovative new ways to improve our users’ listening experiences to meet their needs.”

When I pressed the play button, I was greeted by the disembodied and mechanized voice of Spotify’s Head of Cultural Partnerships Xavier “X” Jernigan, the model on whom the technology was trained. “Thanks for showin’ up,” he says, in a way that makes me particularly aware of the fact that he dropped the g on showin’. After informing me that he is an AI, he tells me that first, he is going to play some artists with a “similar vibe” for me. When I tire of the songs he’s curated (in this case, mostly 90’s Bristol trip-hop), there is a small button I can press at the bottom of the screen, ordering the DJ to change the music to something more suitable for my mood. 

“Okay, let me change it up,” he says, and I feel my initial reaction of visceral discomfort intensify. “I’m gonna hone in on a genre for a little while. Got a few quintessential indie songs for you. First up, something from Oberhofer.”

For a while, I listen to a few AI-selected tracks, interspersed with occasional commentary from the “DJ” about the history of the artist, album, or song itself. All the while, there is an uncanny knot in the pit of my stomach that refuses to unravel.

Boone Ashworth captured my feeling of unease in an article he wrote for Wired. “The interruptions never feel warm or personable,” he writes. “You may hate when a dipshit human shock jock word-vomits over the outro of your favorite song to tee up an ad break, but at least there is indeed a dipshit human behind that action.” 

With the increasing popularity of AI-generated art and culture, the notion of having a human––even a dipshit one––behind any given act of creation is no longer a given. Take the field of visual art, for instance, where it is becoming increasingly difficult to distinguish between art generated by artificial intelligence versus that created by a living, breathing human being. If the two are indistinguishable, quality cannot be the sole factor by which we judge art. We do care about the fact that what we consume is created and curated by a human we have just never before been faced with the possibility of an alternative. Only now that such an alternative exists, we are forced to face that which we have taken for granted, left with an inarticulable cry for humanity. 

“We are listening in the time of the cloud.” The opening sentence of Ben Ratliff’s book Every Song Ever puts it succinctly but cuts to the core of contemporary music consumption. Everything is digitized, algorithm-ized, and packaged just for us. Media companies constantly keep track of our streams, allowing us access to analytics that order and organize what we already like, and point us in the direction of new things we might also enjoy. From here, a thing like Spotify’s DJ can begin to think about personalization.

 “With these advances,” Ratliff continues, “we can essentially be fed our favorite meal repeatedly. We develop a relationship of trust––with what? Whom? A team of programmers? Our own tastes, whatever that means, translated into a data profile?”

According to Ratliff, cloud-listening positions us in an echo chamber of our own creation. But even in making such a bold claim, he strategically refuses to define taste. Taste feels both tangential and quintessential to the questions of personalization and personhood at hand. 

I believe taste sits at the core of this phenomenon to the extent that our tastes are becoming just another facet of our increasingly hyper-individualized identities. If we can find a music genre somewhere between Japanese shoegaze and industrial hip-hop, somehow we feel as if we can differentiate ourselves, even marginally, from those in our periphery. 

Cultural critic and essayist Rayne Fisher-Quann writes, in a short essay entitled micro individuality, “everyone is jostling for attention in a crowded room, struggling to differentiate themselves within an algorithm that exists to turn their personhood into a commodity, subverting and subverting again and re-subverting and de-subverting until they’re right back in the mainstream.” Fisher-Quann’s articulation of this paradigm points us to something profound: the omnipresence of marketing tactics that seek to commodify the cultural consumer. If an algorithm trained on our listening habits can tell us that we are unique, then goddamn it, we want it to be right. This is the linchpin of the human-algorithm trust that Ratliff brought into question. 

Yet as much as we may want to believe in it, trust between person and algorithm will always be built on wavering ground. When the novelty of being told in DJ X’s smooth male croon that our taste is entirely unique wears off, we are left with the daunting uncanniness of the whole enterprise. This doesn’t make me feel special, only scared.

Leave a Reply