An indie musician’s song got billions of streams. Now he’s worried about Gen AI music.
Streaming upended the economics of music. Now musicians are grappling with how their sounds and lyrics may be ripped off by AI.
For Shane Culloty, an Irish folk musician known as Winter Aid, making a living as a full-time musician was never abundantly realistic, as is the case for many independent musicians without a label or promotion vehicle.
That is until his song, The Wisp Sings, was included in a Spotify playlist called “The Most Beautiful Songs In The World.” From there, Culloty’s song snowballed to even more Spotify playlists, including playlists of songs to be depressed to, songs to lay awake at night to, or songs to cry to. All of a sudden, Winter Aid had a following. A song that previously had a few hundred streams and no promotion behind it was now seeing millions of listens. Today, The Wisp Sings has seen over 225 million streams on Spotify.
But that wasn’t all: during the pandemic, The Wisp Sings also went viral on TikTok. It became a meme sound for heartwrenching or heartwarming tearjerker videos, videos of spouses finding out about infidelity or grandparents receiving a phone call from a grandchild. The song has been streamed billions of times on TikTok, leading to HBO picking the song for a show’s emotional high point and Amazon picking the song for a show’s emotional low point. This virality has enabled Culloty to do what very few can: pursue his art and his music career full-time.
Despite having more of a deeper understanding of tech policy than most musicians–Culloty has a Masters in Public Affairs from UC Berkeley, previously worked in policy security at Google and YouTube, and currently is an Ethics Fellow at Stanford–he is still not entirely sure as to how his song was proliferated by the algorithm. He was able to pinpoint one significant TikTok account affiliated with the Biden campaign that used the sound from his song. But other than that, he had the surreal experience of sitting back and watching the numbers skyrocket.
Even more confusing is that Culloty always saw the song as a “reasonably happy” song, not a tear-inducing one. In an interview with Irish news outlet RTE, Culloty connects The Wisp Sings to his “voice, piano, guitar - the sound of a happy home, the drafty old Dun Laoghaire apartment I shared with the woman I love.” He says you can hear his fiancee washing dishes in the background, while lyrics include “imagery from the folklore of my Kerry childhood with abstract expressions of domestic contentment.”
Culloty is aware that his trajectory as a musician is far from common–as he told me, “there are so many amazing bands that never get heard.”
According to More Perfect Union, it now takes 1,000 streams for an artist to afford a cup of coffee, while musicians have to get more than 800,000 monthly streams to make as much as a $15 per hour full-time job. While Spotify made over $1.3 billion in profits last year, musicians make tiny fractions of a penny per stream.
What complicates these economics even further is the ease with which music can be created or ripped off through generative AI, a new development that concerns Culloty and his community of musician peers. An analysis from NPR found that “People are already using AI models to analyze artists’ signature songwriting styles, vocal sounds or production aesthetics, and create new work that mirrors their old stuff without their say.”
Now, Culloty told me, musicians will have to not only invent beautiful pieces of music and game algorithms and platforms, but fend off theft of their original music. Music from humans is at risk of being more easily chopped up or copied–with a focus on distribution systems rather than what’s musically compelling or from the soul. Music, as it becomes cheaper and faster to make, may become inherently “cheaper.”
Culloty does not know yet whether his music has been used yet, but he imagines it’s pretty likely.
“It’s unsettling to listen to Discover Me on Spotify, or music at a cafe or gym, and doubt whether the music is from the voice of a human,” Culloty told me. “In some cases, it will be harder and harder to discern which is which. And musicians will be forced to compete against cheaper content.”
The AI music industry is growing quickly, though we may not even be consciously aware of it. In a Morgan Stanley report from January, an analyst shared results that Gen Z and millennials listen up to three hours of Gen AI music every week, being ingested mainly from YouTube and TikTok. YouTube today just unveiled an AI playlist generator. And platforms called Udio, Suno and Klay allow for users to type in prompts such as “create [music] with an artist’s voice and style.”
While big labels like Universal Music Group (UMG) and Warner Music Group (WMG) initially sued these Udio and Suno, they swiftly not only settled out of court, but partnered with the platforms–claiming this was done for the “protection of the rights of our artists and songwriters” and to fuel “new creative and commercial possibilities.”
For those concerned about the negation of creativity–should everyone be able to make music, or does that dilute the human ingeniousness, talent, and soul behind real music? –there have been policy measures and hard lines being drawn in the sand. Bandcamp put out a statement in mid-January called “Keeping Bandcamp Human,” stating that “musicians are more than mere producers of sound.”
Further, they clarify that: “Music and audio that is generated wholly or in substantial part by AI is not permitted on Bandcamp,” and “ Any use of AI tools to impersonate other artists or styles is strictly prohibited in accordance with our existing policies prohibiting impersonation and intellectual property infringement.”
And there have been some legislative acts to preserve artist livelihoods, like the ELVIS Act in Tennessee–which stands for Ensuring Likeness Voice and Image Security, and protects from unauthorized use of musicians’ name or likeness. Congresswoman Rashida Tlaib (D-MI) also introduced the Living Wage for Musicians Act, in collaboration with the United Musicians and Allied Workers (UMAW), which would tax streaming companies and generate a new streaming royalty to compensate artists more fairly. And in Ireland, the government has committed to making around $1,500 per month of basic income for 2,000 artists permanent.
Some say that because AI is inherently derivative, it will only draw from past music, and therefore musicians shouldn’t be fearful as they embark on creating new works. And a few musicians like Imogen Heap and Grimes have said that AI is something to be used–either for musicians’ own creative processes, or to enable fans to make offshoot music and split the proceeds with the original musician.
But for many lesser-known and independent artists, the idea of how copyrights and musical originality are honored or enforced in the face of generative AI remains an open question.
“I worry that the incentives don’t line up all that well to try to make a living,” Culloty told me. “The stakes are high, and it’s a good time to speak up for songwriters and creators who are not represented by major labels or other large entities.”