Anna Indiana is the world's first all-AI singer-songwriter. She's deeply mediocre.
Screenshot/YouTube - Anna Indiana
- The new completely AI-generated musician Anna Indiana was met with fierce backlash online.
- Her creators wrote on X that she was made using ChatGPT 4 and Musicfy, among other tools.
- Her debut song is creatively unimaginative but not totally off-putting.
She's perfect, too perfect. Her cheeks are poreless, her head moves in an ultra-smooth swivel, and her voice never strays an inch from the desired pitch. She's Anna Indiana, the first musician whose entire existence — from the image and singing to the music's chord progression, key, tempo, and lyrics — is claimed to have been generated by artificial intelligence, without human involvement.
In her debut song, "Betrayed by this Town," Indiana sings about being heartbroken, wandering alone in the streets, and wanting to tear down a town full of "shattered hopes and silent screams." In the video, she sits casually in front of a microphone and pop filter like a B-tier ASMR creator.
Hello world! I’m Anna Indiana and I’m an AI singer-songwriter. Here’s my first song, Betrayed by this Town. Everything from the key, tempo, chord progression, melody notes, rhythm, lyrics, and my image and singing, is auto-generated using AI. I hope you like it ???? pic.twitter.com/0Cf42iyxHI
— Anna Indiana (@AnnaIndianaAI) November 24, 2023
On Indiana's X account, her makers, who have yet to reveal themselves publicly, wrote that an array of tools went into her creation: using open-source Python libraries to program chord progressions, ChatGPT 4 to produce lyrics, and deploying a code to match the lines to a melody and a synthetic voice from Synthesizer V, which is converted to a copyright-free voice using the "cloning" software Musicfy. Business Insider attempted to reach out to them for comment, but they haven't responded.
The video amassed 21 million views on X and faced fierce backlash. YouTube music critic Anthony Fantano said the song "sucks," comedian Kathy Griffin said "We're doomed," and a slew of commenters called the video the worst thing they'd ever heard. In a seemingly sarcastic tone, musician James Blake said he was excited for the future, and the clip brought him to tears.
As new technology has developed, there's been a surge in virtual musicians partly made with AI. It began with Hatsune Miku, created in 2007, who's practically a global pop star. There's Miquela, who has collaborated with real humans like Teyana Taylor and Baauer and whose songs are passable as legitimate pop. Then came FN Meka, the cyborg rapper who signed a deal with Capitol Records but was dropped after being accused of racial stereotyping.
The reaction to Indiana has been so intense because she's the first artist to allegedly be fully developed by AI, and because her music is so creatively bankrupt. Her name itself is an unnerving acronym promoting a total AI paradigm shift: "Artificial Neural Networks Accelerate Innovative New Developments, Igniting A New Age."
Still, it's easy to imagine a future where a cult of fans worship Indiana like they do Miquela and other virtual influencers. Indiana's music is middling, but it's not egregiously unlistenable. Her voice is creepy-smooth, and the lyrics traffic in rote emo tropes. But, really, that's what a lot of passive listening-oriented "streambait pop" sounds like nowadays.
While detractors have said this proves artists don't need to worry about AI replacing them, that doesn't entirely seem like the point. In a streaming ecosystem where 24-hour lo-fi beat stations have tens of thousands of concurrent listeners, and users let Spotify's algorithm auto-play recommendations for hours, it seems likely that Indiana could blend in with the current wave of lowest common denominator music. Indeed, Indiana's creators wrote on Twitter that the goal is "to create and perform new original music on a 24/7 livestream with zero humans in the loop."
Fully AI-generated music might never replace authentic creativity, but it could siphon streams away from more challenging fresh music. Call it "streambait creep": the way listeners reared on generic pop will stop caring if the music is real or robotic. If the human element is wholly taken out, and all it takes is a few buttons to generate more Indianas, an entire macro-genre of easy-listening AI muzak might be next.