People Cannot Stop Dunking on that Uncanny “AI Singer-Songwriter”

Someone decided it was a good idea to make an entirely AI-generated “singer-songwriter,” and she’s so uncanny and unintentionally hilarious that people have begun making their own renditions of her “coffee shop vibes.”

Named “Anna Indiana,” which is apparently a tortuous acronym for “Artificial Neural Networks Accelerate Innovative New Developments, Igniting A New Age,” the creepy-yet-basic singer-songwriter avatar was crafted in the shape of a young woman by unidentified creators.

In a post debuting the avatar’s second single, titled “The First Step,” those creators said (in Anna’s voice) that they used several AI tools to make her, including OpenAI’s DALL-E image generator and GPT-4 large language network (LLM), Adobe Photoshop’s new generative capabilities, and the royalty-free AI music generator Musicfy.

In the videos for both singles, the freakishly smooth-skinned and large-headed avatar is animated only from the neck up, and her hair doesn’t move as the face twists within the boundaries of itself. It would be funny if it weren’t so sketchy, and it makes one long for the days when Hatsune Mikuwas the world’s buzziest virtual artist.

Like her first song “Betrayed By This Town,” the lyrics sound like they were stolen from the diary of a 13-year-old girl — if that girl just so happened to be a robot, that is.

So hilariously stereotypical are her lyrics, in fact, that people have begun using AI tools to make their own satirical versions of Anna Indiana, going so far as to name their avatar “Connie Connecticut” — complete with pointedly AI-butchered fingers — and to program her to sing goofy tweets in her “coffee shop vibes” aesthetic.

In one video, the even more robotic-sounding dupe sings a song titled “Stupid Boys,” but the lyrics, as a background image behind the avatar illustrates, are actually a viral and since-deleted tweet from 2015 clearly intended to poke fun at such rote #justgirlythings tropes.

Podcaster Charles Austin, the brain genius behind Connie Connecticut, has kept up the bit and photoshopped the same generic cozy cafe girl image at the lighting of the Christmas tree at Manhattan’s 30 Rockefeller Plaza and at a Taylor Swift concert, where the avatar, disposable coffee cup in hand, is superimposed over a crowd with the real-life singer catwalking downstage in the background.

“Having so much fun at the Taylor Swift concert!!” Austin tweeted in the avatar’s voice. “My deep learning algorithm leads me to conclude that basically she is ‘mother’!”

The dunks, of course, don’t end there. In one particularly goofy throwback, someone left in the opening to Anna Indiana’s first video and then, once the music came in, played Rick Astley’s “Never Gonna Give You Up,” an incredible 2023 rickroll if we’ve ever seen one.

Real-life drummer Damon Krukowski of the legendary dreampop band Galaxie 500, meanwhile, made a salient point about the content of the AI singer-songwriter’s first song, which includes in its chorus that she wants to “burn it all down.”

“Uh that AI singer–songwriter launched with a bitter song about betrayal and collective retribution,” Krukowski tweeted, “maybe this is actually significant…?”

Let’s be clear: podcasters and musicians dunking on something doesn’t mean anything on its own, but given how objectively bizarre and bad Anna Indiana’s quote-unquote music is, we do, in this instance, have to hand it to them.

More on AI music: YouTube Launching AI Tool That Clones Singers’ Voices