
I was recently speaking with a group of fellow musicians about AI-generated music appearing on Spotify. I argued that such music, not “AI-enhanced” or “AI-assisted” music, but music created start-to-finish with a prompt (or a few) fed to an app like Suno, should be against the terms of service of a platform like Spotify. At the very least, such music should not be monetizable. But several people disagreed with me, saying that modern pop music is already so reliant on digital processing tools (like autotune) that using AI for an entire song isn’t unthinkable.
I think this is wrong for a few reasons, but it actually led me to another epiphany about other musicians’ and music appreciators’ perception of “pop” music: that is, that pop music today is mostly already technology-generated.
I listen to what I would consider “pop” artists like Chappell Roan and John Mayer, but I rarely listen to what plays on Top 40 pop radio stations. When I do, I usually regret it. Such stations seem to be populated only by music crafted purely to make it onto the top charts; the songs often contain predictable production and composition, as well as a lack of feeling. Morgan Wallen’s cover of Jason Isbell’s “Cover Me Up,” for example, received nearly 600 million streams compared to the original version’s 80 million—but Wallen removed the original emotion of the song. Isbell’s original vocal performance is raw, scratchy. Wallen’s, however, is overcompressed, autotuned. And Isbell’s original simple production, involving only a few instruments, is replaced by a wall of stereotypical country music noises.
But this example is a bug, not a feature. In contrast, both Mayer and Roan not only write their own lyrics and accompany themselves with real instruments, but are active participants in the production of some of the most popular songs of the past decade: Mayer and Roan collectively account for more than 52 million monthly listens on Spotify alone. So where does this perception that pop music is no different than AI “slop” come from?
I grew up without streaming platforms like Spotify or Apple Music. The music I listened to was the music I got from friends or family, or music that would play on the car radio. To this day, when I incidentally run into music playing on radio stations or inside a mall, I am still amazed to hear songs that I thought were overplayed when I was in middle school. Yes, that is the definition of popularity. Yes, the role of a radio DJ is different from your friend from college who makes great yacht rock playlists. But the quality of Top 40 stations’ playlists still disappoints me. There is great music out there and it just doesn’t make it to mainstream radio stations. A24 movies are certainly made for a different crowd than the crowd lining up for DC movies, but at least A24 movies still get theatrical releases at AMC.
So it doesn’t surprise me that many people have a poor taste for pop music. They’ve been listening to slight variations of the same songs for more than a decade, and the songs that replace them are often perfectly engineered copies of all of the musical structure and energy of the tracks they replace. But pop music, music that has mainstream appeal and (at least in the past decade) contains a danceable rhythm often associated with hip hop, is associated with some of the most inventive artists of the past decade. Without AI.
Watch Dan Nigro—the producer of Chappell Roan’s “Good Luck Babe” and her album The Rise and Fall of a Midwest Princess—explain how he gets the most out of Roan’s incredible voice. Nigro is actively making decisions about the subtlety of her volume, pitch, and timbre from track to track. By the third chorus of the song, Nigro uses 13 separate tracks of Roan’s voice. The song has 1.8 billion streams on Spotify, not counting listens on any other platform.
Watch John Mayer explain how he got Pino Palladino—legendary bass player for Mayer and other artists like Jeff Beck, Don Henley, and David Gilmour—to record one of the cleanest bass turnarounds in “New Light.” The song Palladino is playing on is unapologetically “pop.” The lyrics are simple and the production is subtle, but there is undeniable human artistry at work. Mayer didn’t ask ChatGPT how to make his turnaround flow more naturally: Instead, one of the most legendary bass players of all time contributed his creativity, and the song is better off for it. Now the song has more than 800 million streams on Spotify.
You may not like the songs or artists, but there is real undeniable artistry appearing in popular music. More importantly, there is real undeniable human artistry. A few tracks may be autotuned or pitch-corrected, but humans are lifting the serious creative weight of songs.
I find AI-generated “slop” like overweight people pole-vaulting and bunnies jumping on a trampoline entertaining and relatively harmless. But I don’t want to consume music that was the dystopian result of some basement-dweller or consultant’s interaction with an AI tool.
Have you ever had an initial reaction to a piece of art, then had your perception changed when you learned the human story behind it? B.B. King’s songs aren’t just enjoyable tunes; the music is made better by its context and the story of the artist. King, a black man, grew up in the Deep South in the 20s and 30s (explaining at one point that “where I came from they used to hang them every week.”) and became one of the most influential musicians in America’s history. Many would be disappointed if I informed you that Eric Clapton’s “Tears in Heaven” was the product not of the tragic story of the death of Clapton’s 4-year-old son, but the product of the following prompt: “Write me a sad song about the death of a father’s young son. Make it in the style of Jimi Hendrix mixed with Muddy Waters. Oh, and add some pedal steel.”
This isn’t to say that the growing presence of digital studio shortcuts in music isn’t worrying. But there’s no shortage of extremely popular human artists making quality music from scratch. And we should make some effort to keep it that way. Using digital studio processing to clean up an accidental sharp note doesn’t automatically make a song engineered slop for some Top 40 list. But what happens when a few overly giddy “innovators” at Universal Music Group start pumping the music market with thousands of songs created start-to-finish via prompts fed to an AI music generator? I think something awaits that is far more depressing and sinister than autotune.
















