A.I. Panic Hits Music City

On a recent Tuesday evening in Nashville, I stood for an hour in a long line on the front porch of the Nashville Songwriters Association building on Music Row, along with about a hundred other would-be songwriters. Some had been waiting as long as two and a half hours, bundled against the cold. Only the first fifty in line would earn one of the coveted spots to play a song for a music publisher. A girl with a clipboard eventually made her way down the line, handing out numbered cards as those of us further back anxiously waited to see if we'd make the cutoff. Close call. I was number 49.

Inside, we crammed into chairs lining an out-of-use recording studio and waited for a young, fresh-out-of-college A&R girl from the publisher to enter the room. Once she arrived, she'd have the unenviable task of listening to the first ninety seconds of each song and telling hopeful songwriters whether she was interested in "taking" a copy. In about two hours, she would "take" perhaps half a dozen out of fifty, which means nothing more than that she will (maybe) listen to the entire song at some point. The first verse and chorus is all the time you get to impress her. Speaking from experience, the most likely outcome of having your song "taken" is that you will never hear from her again.

Playing your songs in front of a room full of peers is a little like being naked in front of quasi-strangers. Most of the time the singing isn't great; the sound quality is bad. Nevertheless, these are songs people labored over, and there's something beautiful about the vulnerability of the situation. The room is full of moonlighting musicians like me—dads who work regular jobs, girls who grew up with Taylor Swift posters on their walls, retired vets with overgrown beards, hippie grandmas who know only four chords on the guitar. The songs range from good enough that you wouldn't be surprised to hear them on the radio, to as amateurish and cringe-worthy as the worst performance you ever heard on American Idol. But everyone is happy to be in the room, and when a good song starts playing, everyone nods in respectful appreciation.

This world runs on faith that every one of us is just one great idea, one great hook, one lucky break away from the holy grail: getting your song cut by a major recording artist. We show up and wait and jockey and endure public humiliation, and we do it gladly. We are on the far outskirts of the music industry, what Hunter S. Thompson described as "a long plastic hallway..." Far from the gleam, fame, and fortune of show business. We're in the minor leagues, at best, but we're still in the game.

Then, a dozen or so songs into the pitch meeting, something different happens. A catchy groove kicks in under a smoking guitar riff. The track sounds like it was cut by top professional session players. And then, as the vocals kick in, it goes to another level. Powerful, soulful. Everyone's ears perk up. But as the tune unfolds, a strange feeling takes over. The music and the singing are world-class, but the melody and the lyrics definitely are not. What we have is a great recording of a bad song. A guy behind me mutters, "I could tell in about ten seconds that it was Suno."

Suno is the AI music app sweeping the music world and raising serious questions about the future of the industry. By typing a few prompts into a text box, you—or rather, the algorithm—can generate professional-sounding songs. If you upload your own melodies or lyrics, it can create new versions of your original content. As one song after another plays in the pitch meeting, more of these AI-produced tracks appear, most carrying that same disorienting quality: songs that sound better than they actually are, polished to a high gloss by machine production.

As Barack Obama once said, you can put lipstick on a pig and it's still a pig. Likewise, you can have Michelangelo paint a mural on the ceiling of your uncle Ned's double-wide trailer, but that doesn't make it the Sistine Chapel. Whatever one thinks of the songs, one thing is certain: AI is alarmingly good at producing music at a quality level that, until now, required highly-paid professionals who had spent their lives honing their craft.

Songwriting, Nashville style, is a craft, and a slow-burning one at that. You spend years learning how to write a great bridge, how to make a hook land, how to fit your whole life into three minutes and fifteen seconds. I had heard friends in the business grumbling about what AI was doing to all of that. But I knew I had to find out firsthand. So I went home and fed my song into the machine.

***

Confronting AI for the first time as a musician can be harrowing. I'd recently watched a video posted by a local touring guitarist—the kind of sideman Nashville produces in abundance—who described receiving an AI guitar track from a client as a reference. The problem: the AI-performed track was so good, he wasn't sure he could match it. He felt there might be only five human beings in the world who could play it so well. This is coming from a professional in a city that represents perhaps the greatest concentration of musical talent in the world.

So it was with some dread that I picked a pop-country tune I'd written recently and uploaded it to Suno, instructing it to “Make a dark, soulful indie country cover of this song with a driving beat.” The platform can generate a complete song from scratch. Meaning, a person with zero talent and almost zero effort can make songs. People are currently creating more than 7 million songs per day on that platform alone, enough volume to surpass the entire Spotify catalog every two weeks. But the technology can also be used more subtly, to create faithful covers of work you've already recorded. It's like having your own band of professional musicians in a box, ready to take direction. That's what I chose to do: I uploaded my rough demo, along with my lyrics, asked it to make a faithful cover, and clicked "generate."

The results were all over the place. One early experiment produced something that sounded like a hit single. The AI singer was so soulful that I would have pulled over to Google him had I heard the track on the radio. But the same version had a hideous, wildly inappropriate drum sound that started firing in verse two, as if Christopher Walken had possessed the algorithm and started demanding more snare. Other versions produced strange discordant moments or went emotionally flat. Working effectively with AI in music isn't a single push of a button. It's an iterative process of trial and error. But after making many versions, I found two keepers.

You can download your Suno creations with all the tracks separated into individual instrumental tracks. So you can mix and match takes, keep some instruments and discard others. I pulled everything into my home recording setup and combined the two best versions into a single track that floored me. Not long ago, producing something that polished would have cost me thousands of dollars and a couple of days with a producer, session musicians, and audio engineers.

There was only one problem: the AI singer had possibly sung my song better than I ever could. In a town full of great singers, I'm considered at least a good one. But the AI voice had everything—incredible control, emotional nuance, and a back-country drawl I'll never have as a Florida native. I suddenly knew exactly what that guitarist had felt when his client sent him that extraordinary AI reference track.

Still, the AI offered something genuinely useful: production ideas I wouldn't have thought of myself. A minor chord in the bridge that improved the arrangement. Taking the final chorus up an octave for power and exuberance. A cool twist on the background vocals in the outro. Killer mandolin parts injecting energy into the choruses. These are the kinds of ideas I might have paid a top producer to provide. I was getting them for the price of a $15 subscription.

But I'm not ready to give up my voice. So instead of surrendering the microphone, I started collaborating. I muted the AI singer's track and recorded my own vocals over the AI instrumentation. And my take is clearly better for having heard the AI's arrangement first.

Was this collaboration successful? Judge for yourself. Press play below to listen, and learn how I used AI to create a professional-level recording on one of my songs:

There's no question that AI helped me produce something more polished than I could have managed on my own. And the price, compared to what it would have cost with human professionals, was essentially nothing. For reasons that are as much economic as creative, I predict that AI will soon be a part of the production of most commercial music you hear. Is the end product less "human"? That's for every listener to judge.

***

My friend Justin is among Nashville's elite guitar players and has performed on albums and stages with some of the biggest names in music. Whenever we work together, he adds ideas and textures that never would have occurred to me on my own, and the result is always greater than the sum of its parts. On a recent afternoon at his house, he played guitar and mandolin on a new “human made” demo while we laughed and shared coffee. It was a song that, like my friendship with Justin, means a great deal to me. In moments like that, it's impossible to separate the music from the people you make it with.

That element—genuine human collaboration—is what's most at risk from what's coming. If I were to feed the song Justin and I recorded together into an AI model, could it play his guitar parts better than he did? I don't know, and I'm not sure I want to find out. AI collaboration that replaces human collaboration is a net loss in my mind. On the other hand, when it enhances work I'd otherwise be doing alone, it's a win. But clearly it’s capable of replacing human talent. Musicians, audio engineers, and songwriters are all worried about what's coming, and rightly so.

The music industry is playing catch-up. Companies like Suno trained their models on copyrighted material, meaning every song you generate involves a kind of meta-sampling of a vast library of existing music. All three major U.S. music companies have filed suit against Suno. Warner Music has since reached a settlement under which artists it represents will begin to receive some compensation. But the broader legal framework remains unsettled, and the technology is advancing well ahead of lawmakers' ability to respond. Some musicians are going to lose their jobs, and the cruel irony is that they may be replaced by AI trained on their own work. On the other hand, a talented teenager somewhere in her bedroom will use these tools to write and produce extraordinary songs that touch millions of people who would never have heard of her otherwise. That's the genuine promise. The question for guys like Justin is: at what cost?

***

Imagine I play you a song and it moves you—makes you feel something, makes you dance, makes you cry or smile. Now imagine I tell you it was 100% AI-generated. Does it matter? I think it does. When I write a song, there are pieces of my life in the lyrics, real human breath behind every word. Art disconnected from human experience loses much of its power to move us. A robot might carve a block of marble, but a great sculpture compels us precisely because it is so extraordinary that a human produced it. Michelangelo's Sistine Chapel staggers us not just for its visual form, but because we understand what an extraordinary talent—and effort—its execution required.

Music is more than soundwaves. It helps us connect to other people. In moments of profound joy or grief, we reach for it to help us make sense of what we're feeling, to express what language alone cannot. "That's when art's not a luxury," as the actor Ethan Hawke puts it. "It's actually sustenance. We need it." Somewhere a heartbroken teenager is in her room crying over a breakup. An overworked mom is stuck in traffic in her minivan. A couple is having their first tentative dance. Music has the power to speak to all of them, where they are.

Aristotle understood imitation—mimesis—as the basis of art: the creative process can produce something new that takes on a life of its own. But what happens when that act of imitation is performed by a machine, over and over, endlessly? Will the art of the future be nothing more than a sterilized copy of a copy of a copy?

It is said that when God created the universe, he did so with words, and that even the stars make sounds. When humans create, it’s then that we come closest to our divine nature. When breath moves through a saxophone reed or through vocal cords, when words rise from real human memories and land on the lyric page, something like love passes from artist to listener. The meaning of that something is tied to the common human experience they share. This connection is sacred and mysterious.

***

My experiment has left me in genuine awe of what AI music tools can do. Empowering talented young artists who otherwise lack the resources to record is a wonderful thing. But we are about to be awash in a sea of musical slop—hundreds of millions of songs generated without effort or forethought. Those of us who are creating, whether on grand stages or in our bedrooms, must choose to use these tools in ways that expand human creativity rather than replace it. Music is made from the souls of the people who made it. People need to hear the tremor of real voices, imperfections and all. Yet I know that every time I upload a melody or lyric I've written, I'm training the algorithm to mimic the best of what I can do—perhaps to the benefit of some stranger who will do nothing more than push a button.

A few nights ago, I carried my acoustic guitar to a bar on Nashville's west end to play one of the longest-running songwriters' rounds in town. Three songwriters take the stage at a time and take turns playing their originals. It's a universe away from AI music: the clink of beer bottles on tables, nothing but a few feet of open air between you and the audience. And there, briefly, I did something AI can't do. I forgot the first line of my song. Only to remember it again and share it with a room full of strangers. For a few minutes, I was singing and everyone in the room was listening. In that moment, I was doing something no algorithm can replicate, sharing part of myself. Every song I write has a piece of me in it. No AI can say that.