Artificial intelligence has leaped from the nerdiest of technological fringes to the mass market of pop culture memes. The classic dystopian worry about the rise of murderous machines has been supplanted by a more immediate and consumerist question: Can AI entertain me? The answer, apparently, is yes. In this moment of future-be-damned freewheeling, we’ve tapped the immense potential of generative AI and machine learning for a wide range of novelties, from writing a TV commercial to de-aging Indiana Jones to putting a puffy jacket on the Pope.
Spotify, the music streaming giant and power player in the 21st-century music industry, has joined the fray with its own AI offering. DJ uses generative AI to create a voice that speaks directly to individual Spotify users while serving up a highly personalized playlist based on the user’s tastes and history.
In a friendly voice that goes by the name X, Spotify’s DJ can tell a user that it’s going to play a specific genre of music, or resurface some of that user’s most played songs from three winters ago, or announce a new song from a band that the user hasn’t heard but is likely to enjoy.
When DJ launched in February as a beta release for premium customers in the U.S. and Canada, it was quickly showered with lustful anticipation from outside the test geography. “When will it be available for India??” asked one YouTube user. “BRING IT TO EUROPE!!!!” demanded another. The world has been shown the rare new AI feature that connects to an already familiar app that many millions have come to use on a daily basis. The power of artificial intelligence is now making playlists.
But Spotify sees DJ as much more than an AI flash in an algorithmically created pan; according to the company, it’s an essential tool for helping people find new music they might actually like. When so much of the internet has become an ungraspable fog of algorithmic curation, DJ offers a window of transparency into its recommendation process.
“I think . . . music [discovery] is a core human need,” says Emily Galloway, senior product design director for Spotify’s Personalization team. “I don’t think that core human need has changed over time, but we continue to try different types of technology to solve for that and innovate in that space.”
Helping people find new music is also key to Spotify’s business model, giving them more reasons to keep paying to listen. AI, it turns out, is a handy way to make that happen.
But DJ goes beyond scouring user histories and suggesting new songs to play, which Spotify already does. The true power of the AI is its voice element, which is based on the playful voice of Spotify’s head of cultural partnerships, Xavier “X” Jernigan, who may be familiar to some Spotify users as one of the hosts of its former morning radio show The Get Up. DJ X speaks the listener’s name, announces some songs, and provides factoids or background for why certain genres or artists are being played, directly though listeners’ headphones and speakers.
“It’s really bringing context to your ears because not everyone’s listening [and] looking at their phones,” says Galloway. “We’ve found it to be really successful because people are more likely to give a song they haven’t heard before a try if they understand why they are being recommended that music.”
Generative AI makes the recommendations and the voice function, but much of the personality comes from a room full of living human writers. Galloway says this team tries to give DJ X personality, and working with Spotify music editors injects the DJ feature with knowledge of music and music culture, all of which trains the algorithm. Which is all sort of like what you might hear from a real DJ on a radio station. “For all of our recommendations and personalization on Spotify, the hope is that it does get better over time,” Galloway says.
Calling an AI tool “DJ” is sure to tick some people off, particularly those actual human disc jockeys who specialize in picking the right song for the right moment. Galloway says naming is always tricky, and Spotify’s DJ feature isn’t trying to replace the human music tastemaker. “What we kept going back to, and what is most unique about this experience compared to most things on Spotify, is that a voice is talking to you,” she says. “We did want the name to nod back to that unique aspect.”
The point of DJ is to streamline and filter through the estimated 100 million songs and tracks in Spotify’s library, helping listeners hear what they like and what they may not know they like. And that also goes for the artists and content makers dumping their creations into what can seem like a black hole. “At the end of the day we’re also trying to optimize for discovery and driving these really deep, meaningful connections. So we have to think about listeners and creators and that matchmaking,” Galloway says. “Personalization really is at the heart of Spotify.”
Recognize your brand’s excellence by applying to this year’s Brands That Matter Awards before the final deadline, June 7.
Sign up for Brands That Matter notifications here.
