By Harmeet Kaur, CNN
On its surface, the saga of FN Meka appeared to be a cautionary tale about the pitfalls of music powered by artificial intelligence.
The digital avatar of a Black man with green braids, face tattoos and a sound reminiscent of 6ix9ine was billed as an “AI-powered robot rapper.” Though the company that created FN Meka said the character was voiced by a human, it insisted that other aspects of the music were based on AI.
Depending on who you asked, FN Meka was either the next frontier in music or a death knell for human artists. FN Meka amassed millions of followers on TikTok, and as the hype around him grew, music industry executives took notice. In August, Capitol Records announced it had signed the virtual character.
Just weeks later, the record company dropped FN Meka from its label after Black music industry professionals criticized the project for using reductive stereotypes. Soon, the narrative began to unravel. The artist Kyle the Hooligan told VICE that he wrote and voiced FN Meka’s early songs, and claimed he still hadn’t been paid. Recently, a co-founder of the company behind FN Meka characterized his previous comments about the use of AI in FN Meka’s music as little more than a marketing gimmick.
The questions raised by the FN Meka case, then, aren’t inherently about the ethics of technology in music. Rather, some experts say, technology is just one layer in larger conversations about cultural appropriation, copyright and ownership.
Virtual artists are on the rise
FN Meka’s digital persona — and the extent to which it was readily embraced by an audience — reflects a cultural shift that is well underway.
From the band Gorillaz to holographic Tupac to Travis Scott’s “Fortnite” concert, artists and engineers have been experimenting with digital alter egos for years. And as the metaverse expands, virtual influencers and avatars likely aren’t going anywhere.
Lateef Garrett, music manager for the virtual record label Spirit Bomb and an industry veteran, says virtual personas open up new creative opportunities for real-life artists. Virtual characters can allow artists to experiment with new musical styles, reach new audiences, and access new revenue streams without necessarily having to be the face of the music. But he says it’s critical that real-life artists have a stake in developing those characters — something that didn’t appear to be the case for FN Meka.
Kyle the Hooligan, who is Black, said he was the original voice of FN Meka and helped shape his sound. He claimed that he was promised equity in the character but was eventually ghosted, telling VICE that “them cutting me out of it was like they basically used me for the culture.” It wasn’t until after he was cut off from the project, he said, that he learned of certain creative choices that have since been critiqued — he also hadn’t realized FN Meka had been signed to a record deal.
“You can’t side step working with Black artists by creating a Black virtual artist,” Garrett says. “Most importantly, you can not replicate the Black experience or Black culture through virtual artists, unless Black people are involved in creating that character.”
FN Meka employed the stereotypical aesthetics of SoundCloud rappers, used the N-word in lyrics and simulated the experience of police brutality and incarceration — a result that, according to some critics, amounted to “digital blackface.” It also highlights concerns about cultural appropriation when creating virtual characters: Is it appropriate for a virtual character to use the slang and markers of Black hip-hop culture if the people behind the character aren’t steeped in that culture themselves? In other words, whose culture is being extracted and who is profiting?
“It’s important that when these virtual artists and AI tech companies are built, the ones making the decisions on character developments and music reflect the community in which they’re trying to reach,” Garrett says. “I think certainly having a diverse staff would solve a lot of the problems that we’re seeing in this situation.”
Virtual characters also create distance between creators and their creation, says Gigi Johnson, who leads the Maremel Institute, a think tank focused on the intersections of creativity and technology. At the height of FN Meka’s popularity, there was little clarity about who exactly was behind the character, confusing consumers and making it easier for developers to evade accountability.
In instances where artificial intelligence is actually generating or assisting music, the questions get murkier.
“Who makes the decision about whether this goes out or gets edited?” Johnson says. “Who stands behind this piece of work?”
AI still has significant limitations
FN Meka’s music may not have been generated by AI. But for better or for worse, artists and researchers have already been experimenting with AI to try and push music to new heights.
In 2020, the company OpenAI released “deepfake” tracks made to sound like the musical stylings of Katy Perry, Frank Sinatra and other well-known artists. The company space150 generated buzz with TravisBott, an artificial neural network that creates original music made to sound like Travis Scott. Meanwhile, the artist Holly Herndon used AI prominently in her 2019 album “Proto,” and has since introduced an instrument that allows users to transform any audio into the singer’s voice.
Even as more artists employ it, music composed by AI is far from perfect. Algorithms are trained to recognize patterns in existing music that is input into computing systems, spitting out a product that’s inherently derivative. In other words, AI — for now — is only as good as the data that feeds it. Even then, as evidenced by many of the nonsensical lyrics generated by TravisBott, a fair amount of human intervention is needed to create cohesive songs.
For these reasons, Nina Eidsheim, a professor of musicology at the UCLA Herb Alpert School of Music, sees AI as yet another tool that musicians will be able to use in creating art — not as a replacement for the artist.
“There are going to be glitches, and there are going to be things that we wouldn’t have imagined,” Eidsheim says. “As much as what the AI technology in itself is creating, what’s interesting is how we as human artists are interpreting it and maybe taking it into our own artmaking — or rejecting it.”
But, as Holly Herndon noted to the music magazine FADER, there are other concerns about ownership and consent when it comes to the data powering these algorithms: What happens when an artist’s work is used without permission to generate music that someone else could profit from? How much of that work can be used before it infringes on intellectual property laws? And how much control does that artist have over their own likeness?
Hip-hop has been wrestling with such questions when it comes to sampling for a long time. But advancements in AI now allow people to go beyond just taking a fragment of another person’s work.
“Is it okay to literally sample someone’s personhood? Are we okay with that, as a society?” Herndon said in the interview with FADER. “And if we’re okay with that, how does that play out within the existing power structures that we already have in society?”
And as more artists are included in the datasets fueling these algorithms, the harder it will become to trace which elements were borrowed from whom, Johnson says.
“If you can’t really take the pie back apart for its ingredients, who should get paid for the whole?” she says.
Tech has long transformed music
Technology has been transforming music, and art more broadly, for generations. And when it comes to AI specifically, other recent endeavors force us to grapple with more profound questions than FN Meka ever did.
Last month, a man won first prize in a state fair’s fine arts competition for an artwork of a space opera scene he generated using AI software, sparking outrage from some artists and resurfacing anxieties that machines are out to replace humans. Though that image was created using the software Midjourney, an app called DALL-E 2 works in a similar way.
The incident also called into question what, precisely, constitutes art: Is there less value in the work if the artist’s only involvement was crafting the prompt that generated the image? When applied to music, does the use of AI diminish the artistic skill once needed to craft songs? Or does it tap into a new skill set and lower the barriers to entry?
While AI systems have come up with close approximations of existing music, and can do so at a scale that surpasses humans, they’re still dependent on real artists to fuel and fine-tune the end result. What is evolving, then, is who gets to be an artist and how artists work.
“Whatever we make, or however we use AI, it’s not going to stand outside any of the things that humans do because we’re involved,” Eidsheim says.
The-CNN-Wire
™ & © 2022 Cable News Network, Inc., a Warner Bros. Discovery Company. All rights reserved.
Discussion about this post