Should Songwriters Use AI to Write Songs?
Whether AI has a place in music is a hot topic amongst creatives. I would like to address this topic from the perspective of songwriting, and specifically, learning to write more powerful songs. The following thoughts are my opinions, and the policy I currently follow within programs I offer and my online community.
Suno is a popular platform through which artificial intelligence can generate lyrics, melody, chords, and rhythm in the shape of a song. It can also dress these elements into full arrangements. It is akin to a collaborator or producer, but with the important difference that it doesn’t react to the song’s direction, meaning and personality, and therefore cannot judge creative ideas, arrangement or production like a human would. It can’t intuit whether one rhythm, melodic pitch, lyric phrase, rhyme, style or approach is more or less powerful than another. It can only show tendencies that, by their very definition, contain the song within a specific genre.
That containment can be exciting at first, and even offer pieces characteristic of the genre that we like and want to continue to use in the final recording. As a tool, Suno can remove some of the need for instrumental or vocal skill, so that those elements no longer have to create a barrier for a listener to truly hear the song. I believe there is great value in this result, particularly for writers who would otherwise not share their work due to lack of access to or affordability of capable collaborators.
In the real world, professional writers are using Suno. I want to make it clear that, though not fully trusted and embraced by writers, it is sometimes used in a spirit of creative curiosity. It is rarely, if ever, used amongst writers and artists setting the creative tone as a decision-maker or equal collaborator. This is what is meant by the term ‘tool.’ As a tool, just like a rhyming dictionary or thesaurus, or a library of our favorite drum samples, it can enable more free creative expression for each individual.
One of the critical issues I see with enabling AI to capture too many song or production decisions is that it reduces our touch-points with music. I, myself, learned to sing by writing songs. I was a piano-player who picked up acoustic guitar my third year at Berklee. My fingers hurt so bad I moved over to electric just to be able to fingerpick. When that felt okay, I moved back to the acoustic until someone said the action on my guitar was too high. Got the action lowered, and tried again. All the while I was muttering vocal lines and lyrics over the only five guitar chords I knew. My songwriting style was being shaped by my growing ability to sing and play. I’d find a new chord, and my fingers felt the shape of it, and my voice found a new pitch for the melody to pull towards.
I didn’t have a country voice, but I wanted to pitch my songs in the country market because I had moved to Nashville with that intention. When I wrote, however, even my country songs weren’t country songs, and that’s because my voice and my guitar playing kept bringing me back to my own sound that was slowly developing. Fast forward to my first cuts, and it’s no wonder the placements I got were with pop crossover artists. That was entirely due to my own personal style, as developed through my physical and mental approach to playing my instrument and singing. Though I wasn’t aware what it was, it was becoming stronger and more unique to me every day. In a world where music can be mass produced, our biggest asset is our personal approach that is a direct expression of our unique voice. It is the one thing that I would never, ever encourage a writer to minimize.
Sometimes songwriters ask me how I can find the good in every writer’s songs. It’s because when I listen, I try to feel what the writer was trying to make. In my body, I know the feeling of a forced rhyme, a missed opportunity with melody or chords, or a groove and vibe that doesn’t feel quite developed yet. I’ve felt all those things about my own writing. The difference is that I’ve also felt when those things peel away and something magical occurs within a song. I’ve gotten to hear ‘yes’ amidst all the ‘no’s,’ and had songs sung by other artists, which is affirmation that I can work with what I’ve got. This is absolutely irreplaceable, and I want you to have it, too.
AI may appear to polish your song. If that helps you to feel more confident sharing at first, I completely understand. But do not let it keep you from integrating the tools into your body, because that feeling when things finally click into place is your ‘taste’ becoming ‘expression.’ When we can seek out real people as collaborators to support us in the areas of instrumentation or vocals, we gain so much more than we do going solo through AI.
So, though I welcome the opportunity to hear your music any way you want to present it, I encourage you to take risks. Share your own raw voice and playing whenever you can, even when it’s imperfect. And if you are a part of our Alumni community, I hope it is a place you feel safe to be truly heard.
Come write with us,
Andrea Stolpe