I (Still) Hope You Never Listen Alone
Revisiting what scares me the most about AI-generated music
Last week, I announced my debut book, Uncharted Territory: What Numbers Tell Us about the Biggest Hit Songs and Ourselves. I was overwhelmed with the positive reception. If you still haven’t, please consider ordering a copy.
One of the last sections of the book deals with artificial intelligence and music. I was reminded of this for two reasons. First, I’ve had to reread the book over the last week while it’s being typeset. Second, Sam Altman — the CEO of OpenAI, the company behind ChatGPT — published an essay on his blog called “The Gentle Singularity.” That essay began quite dramatically: “We are past the event horizon; the takeoff has started. Humanity is close to building digital superintelligence, and at least so far it’s much less weird than it seems like it should be.”
My mind goes many places when I read grand statements about artificial intelligence, especially when those statements come from people who are building it. But one of those places is a sad one. It’s a place filled with loneliness. Given that I am making some final edits to my book, I want to rerun a piece that I wrote about music and artificial intelligence two years ago. It’s a piece that I think was somewhat prescient.
I (Still) Hope You Never Listen Alone
By Chris Dalla Riva
The other day a friend sent me this a New Yorker article from 1978 about the legendary late-night television host Johnny Carson. Here is how acclaimed film director Billy Wilder described Carson in the article:
By the simple law of survival, Carson is the best … He enchants the invalids and the insomniacs as well as the people who have to get up at dawn. He is the Valium and the Nembutal of a nation. No matter what kind of dead-asses are on the show, he has to make them funny and exciting. He has to be their nurse and their surgeon. He has no conceit. He does his work and he comes prepared. If he’s talking to an author, he has read the book. Even his rehearsed routines sound improvised. He’s the cream of middle-class elegance, yet he’s not a mannequin. He has captivated the American bourgeoisie without ever offending the highbrows, and he has never said anything that wasn’t liberal or progressive. Every night, in front of millions of people, he has to do the salto mortale [circus parlance for an aerial somersault performed on the tightrope] What’s more … he does it without a net. No rewrites. No retakes. The jokes must work tonight.
This praise is effusive. And it’s not unique. If you speak to anybody who watched Carson regularly, you’ll hear them heap the same praises on him, albeit less poetically. Anytime I see a clip of Carson, I too am swayed by his charms. But in the back of my head, a thought always creeps in: Maybe everybody thought Johnny Carson was great because there weren’t many other shows to watch late at night.
Compare the late-night entertainment landscape of yore to what we have today. Between YouTube, TikTok, and Netflix alone there is more content than any human could watch in ten lifetimes. This puts us in the weird scenario where we arguably have more great art than ever before but less people to enjoy it with.
What scares me about artificial intelligence is that it will likely exacerbate this situation. This AI-generated recording of Johnny Cash covering the Death Cab for Cutie’s “I Will Follow You into the Dark” is a good example as to why.
First, I want to note that saying this was generated via AI probably doesn’t mean what you think is does. In all likelihood, an artificial intelligence model was trained on Johnny Cash recordings so that someone else’s voice could be mutated to sound like him. A human was still required to sing the song and edit the recording. Nevertheless, here is how this technology could progress.
Play Me a Song: This is where things currently stand. A streaming service knows what songs you enjoy and uses some combination of machine learning to find others you’ll feel the same about.
Combine These Songs for Me: This is what the various “AI” covers proliferating on the web are trying to be. You pick something, like Kanye West singing “Take Me Home, Country Roads,” and it will start playing.
I’ll Combine These Songs for You: Rather than telling the computer what to combine, now the computer knows what you want. This is similar to (1), except rather than having to choose from a set of songs, it can combine things. For example, maybe it knows that you’d love to hear Lana Del Ray perform an acoustic cover of Lady Gaga’s “Bad Romance.”
I’ll Make a Song for You: Now, the computer isn’t working from component artists and songs that you are familiar with. It’s generating a song calibrated to your tastes from scratch. It might have been trained on human-made art, but the output will be distinct from the inputs.
So, why does this worry me? Because you will eventually be listening alone. And I don’t mean blasting your favorite song with ear buds in. I mean that you will be listening to music so finely personalized to your tastes that nobody else will have heard it or will ever hear it. You will be an audience of one.
I’ll admit that I’m being an alarmist. Not only am I assuming that we will surmount all of the technological challenges necessary to arrive at this scenario, but I am also assuming that each person’s tastes are distinct enough where there’d be little to no overlap in AI-creations. The former might not come to pass and the latter might not be true. But the next time you and a friend are both sitting in silence on the couch scrolling through your respective TikTok feeds, you’ll come to realize that pieces of this world already exist.
So, was Johnny Carson really that great? I’m not sure. I write about music, not television. But I don’t think it really matters. What matters is that I’d rather enjoy a decent piece of art with friends than the greatest piece of art by myself.
An Update
In the two years since I published this essay, artificial intelligence technology has progressed dramatically. In fact, much of what I predicted has come to pass. In 2024, a company called Suno went public with technology that allows you to generate recordings from a simple text prompt. Streaming services, like Spotify, now also use artificial intelligence to generate playlists from text prompts.
We have not reached the point where streaming services are generating entire songs in real time based on your tastes. Assuming that you could avoid royalty payments by doing this, I think we will see someone try this at some point. Artists and labels would definitely riot if it happened, but you should be on the lookout for it. Earlier this year, Liz Pelly reported that Spotify was stuffing playlists with artists who didn’t exist. I don’t think we should fear this technology in principle. But we should be weary of it if it’s being used to push us apart more often than pull us together.
A New One
"missing everything" by Amie Blu
2025 - Indie Rock
“missing you,” the latest single from south London singer Amie Blu, makes me want to cry. Though the lyrics alone could make tears well up (i.e., “My days are longer, they feel like three / I’m out of blame anyone but me”), that’s not what’s actually doing it for me. It’s tone. It’s the tone of the palm-muted acoustic guitar. It’s the whispery vocal from the singer. Frankly, Amie Blu could be singing anything on this song, and I’d feel the same way.
An Old One
"All I Wanna Do" by The Beach Boys
1970 - Dream Pop
Last week, we lost Brian Wilson, maybe the greatest songwriter of the 20th century. While I could point to scores of compositions to mourn his genius, I want to turn to a less work: “All I Wanna Do.” Released on The Beach Boys’ 1970 album Sunflower, “All I Wanna Do” is basically a dream pop song decades before the genre was even invented. In fact, if you put this song out today under a pseudonym, I think you could convince people it was a rising indie rock band from Brooklyn.
That is the magic of Brian Wilson. His productions feel like they are from the future, and his melodies exist outside of time. Rest in peace, Brian. It was indeed nice.
If you enjoyed this piece, consider ordering my book Uncharted Territory: What Numbers Tell Us about the Biggest Hit Songs and Ourselves. The book chronicles how I listened to every number one hit in history and used what I learned during the journey to write a data-driven history of popular music from 1958 through today.
Recent Paid Subscriber Interviews: Pitchfork’s Editor-in-Chief • Katy Perry Collaborator • Spotify’s Former Data Guru • Music Supervisor • John Legend Collaborator • Rock Critic • What It’s Like to Go Viral • Adele Collaborator
Recent Newsletters: Pop Music Kill Rate • The Lostwave Story • Blockbuster Nostalgia • Weird Band Names • Two-Hit Wonders • A Frank Sinatra Mystery
Want to hear the music that I make? Check out my latest single “Give It To Me” wherever you stream music.
Listening to music alone… This is a fascinating and simultaneously depressing concept to entertain. I’d like to add some additional perspective.
My grandfather was born in 1885, second generation Irish immigrant. In 1907 at the age of 22 he founded the Caldwell Nursery, a plant nursery, in Caldwell, NJ. I’m 74 now, and it’s impossible for me to count the number times I was there for Thanksgiving dinners, Christmas dinners, anniversary, and birthday celebrations etc.
At one such occasion I remember (quite clearly), my grandfather describing how music was played and shared among people when he was a younger man. He and most of his neighbors were farmers, and occasionally, maybe once every two months, they would all gather together at someone’s house, usually whoever had a piano. Someone would play it while everyone else sang in unison or in harmony or took turns soloing. The women would make sandwiches and serve coffee and - for them - it was very happy and lively evening indeed. Unlike modern TV shows where people stare at their feet in anxious uncertainty while judges might harsher critique them for not “owning the stage“ or whatever, everyone sang for the simple joy of it, and no one was particularly in competition with another.
My grandfather would tell this story often, always with a smile and a lot of affection; it definitely brought back many happy memories for him.
Then his countenance would change into something less than a smile as he recounted how disgusted he was with the invention of the radio, and while all the same neighbors would still come over to each other’s houses every month or two, the dynamic changed, and instead of people singing for fun and playing the piano they would all gather around the radio, sitting in chairs, to now listen to various skits that were being professionally produced as well as professional singers singing. He thought that was a real loss because there was no actual interaction anymore, people were just passively listening to a radio, sitting there in easy chairs, happily, munching a sandwich or smoking a pipe and enjoying the performances. But there was no longer any personal interaction as there had been.
Then his face became sadder still as he talked about how the television had taken the radio 10 steps further in terms of people interacting, or should I say not interacting. The famous architect Frank Lloyd Wright, once quipped “television is chewing gum for the eyes“ and no one believed that more than my grandfather.
My grandfather died in 1972, so I don’t think he lived to see the era whereby almost everyone in the family had their own television in their own room, much less today’s digital world where we all carry the entire globe in our pocket on a mobile phone.
All to say that this is nothing particularly new, and as much as it saddens me, the thought that people might just one day listen to music just by themselves seems to me to be almost a natural extension of what I just described.
Then again, people love to go to shows to watch people perform and that’s something that’s been constant throughout history. So who knows? Thanks for reading this far.
As a fellow musician, born in 1945, I believe a lot of what you’re talking about started in the 80s and 90s of the last century, with programs like harmonizer, and superceeded by the Canadian geniuses who developed band in a box. I experimented with these, and was preparing for a Brass quintet gig, and we needed an arrangement of La Bamba. So I put the melody and the chords into harmonizer, and a generated parts for two trumpets, French Horn, trombone, and tuba. We played it on the gig, and it was a success. But I don’t know who would get the credit for the arrangement. me for pushing the buttons, the programmer, or the community, which built the computer. BIAB has become quite sophisticated. You can enter a chord progression, it will generate a melodic line similar to the improvisations of people like Charlie, Parker, John Coltrane, or Chet Atkins. Then, you can have that melody harmonized in , 3, 4 or five parts, in the style of Glenn Miller, or the style of Gill Evans.
The computers job is much simpler now, because modern pop music is generally so simplistic, all you need is two cords and three notes to make a song, which only needs a generic rhythm, and a lot of processing.
Yours in the future, Andrew Homzy.ca