What Does It Mean to Be a Songwriter with AI in the Room?
How co-writing with a language model is changing the creative process—and why that’s not a bad thing today
As part of my exploration into creative applications of AI—personally and professionally—I’ve been experimenting with what it means to co-write with a language model. Here’s a personal take on songwriting, authenticity, and where AI fits into the process.
In some ways, I think it still comes down to the artist.
AI can assist. It can inspire. It can even surprise you. But ultimately, it’s the songwriter who is shaping the final expression. Their taste, perspective, and intention guide everything—from the direction of the prompt to the decisions they make along the way. They are both the inception and the conductor. So the quality of the final work is still dependent on them.
Can AI hurt the creative process?
Absolutely. ChatGPT might offer a word, analogy, or theme that seems clever in the moment but pulls the song off-course. It can send you down a rabbit hole that sounds interesting but dilutes the emotional core. That’s the risk of using any co-writer, human or machine. It’s up to the songwriter to guide the vision and shape it to their likeness.
If your favorite song was written with ChatGPT, would it still matter?
As a fan, that was the first question I needed to answer.
I consider some of the best songwriters to be magicians. The best songs feel like they’re channeled—like they come from someplace deeper. They transport you. So if someone told me that John Prine didn’t write “Paradise”, I’ll admit: my gut reaction would be, “Wait, that wasn’t just his words?” or “Is this story even his?”
But that instinct is flawed.
We already know some of the greatest songs are written in co-writing rooms—in Nashville, LA, New York—by teams of people. If those same people used ChatGPT to shape a chorus, would that make it less authentic?
What if ChatGPT had existed when Elton John and Bernie Taupin were writing songs? Would Bernie have used it to iterate on a lyric? Would it have made their songs less meaningful?
For an artist like John Prine—where authenticity is central—he simply wouldn’t have written a song that didn’t feel personal, whether he was writing alone or with a machine.
Does it matter where the song comes from?
Sometimes it does. But not always.
Some songs are just meant to pump you up before a night out. Others sit with you for years—like Both Sides Now by Joni Mitchell—because they hold a truth that mirrors your own. In those cases, the backstory behind the song matters.
But even then, it’s still up to the artist. If they’re being genuine, if the words are still true, does it matter whether ChatGPT helped shape the phrasing?
I think one thing people latch onto when they talk about good songwriting is a song they can relate to. That emotional connection is grounded in humanity. And that’s one of the songwriter’s core responsibilities: keep it relatable, keep it human.
When does a song feel like yours?
This is personal—and subjective.
One of my favorite songs I’ve written? I only wrote 10% of the lyrics. But I shaped the chorus, the melody, the energy. It still feels like mine. Because I brought it to life in a way that reflects me.
Other times, I feel like I didn’t “write” the song at all—it just came through me. Like it already existed, and I was just the vessel.
Like Dylan said, "I don't know where the songs come from. If I did, I'd go there more often." Sometimes I’m in the driver’s seat crafting every chord and word. Other times I feel like a passenger, tuning the radio to the right frequency to find a song that already exists out there. Like it's just waiting to be captured and shared.
Some people feel ownership if they wrote the lyrics. Others if they shaped the melody. For me, it’s when I helped steer the emotional truth of the song. That’s when it starts to feel like mine.
What role should LLMs play in songwriting?
Right now, I’m just starting to explore tools like ChatGPT and Claude as co-writing partners. I haven’t found much value yet in writing melodies—but lyrics? creating the right imagery with your words? iterating quickly? That’s where things start to get interesting.
I’ve used it countless times for small tweaks—like swapping a word, finding an alternate descriptor to finish a rhyme, or searching for the perfect metaphor. It helps me get through the bad ideas faster so I can get to the good ones.
I don’t think I’m the best lyricist. So bringing an idea to ChatGPT after I’ve written the melody or structure helps me refine it. It’s not replacing my voice—it’s helping me clarify it.
Isn’t that cheating?
Not to me.
T.S. Eliot once said, “Immature poets imitate; mature poets steal.” He wasn’t condoning plagiarism—he was pointing to something deeper: that art evolves through influence. Through reinterpretation.
I don't see it as steal or cheating—we’re transitioning to a hybrid role where creation and curation are intertwining more and more. Good artists have always been good curators, but now but curation is becoming a core part of authorship, a core part of this changing creative landscape.
Using AI in songwriting doesn’t mean you’ve lost your voice. But it does mean you’re making choices. And those choices still reflect who you are.
Final thought
Songwriting has always evolved with the tools of the time—whether it was the typewriter, the tape recorder, Pro Tools, or now… language models.
What matters most isn’t the tool. It’s the story you’re trying to tell—and whether it still moves someone when they hear it.
I'd love to hear from you.
If you're exploring creativity, songwriting, or AI in your own work—whether you're an artist, a technologist, or somewhere in between—I'd love to know how it's shaping your process.