With nearly 100 million tracks available and more than 600 million subscribers, helping listeners find the music they love has become a navigational challenge for Spotify. It's the promise of personalization and meaningful recommendations that will give its massive catalog more meaning, and that's central to Spotify's mission.
The podcast giant's suite of recommendation tools has evolved over the years: Spotify Home Feed, Discover Weekly, Blend, Dailylist, and Made for You Mixes. In recent years, there has been evidence that this idea is working. According to data released by Spotify at its 2022 Investor Day, artist discoveries each month on Spotify reached 22 billion, up from 10 billion in 2018, “and we're nowhere close to being there yet,” the company said at the time.
Over the past decade or more, Spotify We have invested in artificial intelligence and, in particular, in machine learning. The recently launched AI DJ may be its biggest bet yet that technology will allow subscribers to better personalize listening sessions and discover new music. The AI DJ mimics the radio atmosphere by announcing song names and intros to tracks, something partly intended to help make it easier for listeners to step out of their comfort zones. One weakness of AI algorithms — which can be excellent at giving listeners what they already know they like — is anticipating when you want to step out of that comfort zone.
Combining personalization technology, generative AI, and dynamic AI audio, listeners can tap the DJ button when they want to hear something new, and something less directly derivative of their established likes. Behind the dulcet tones of an AI DJ, there are people, technologists, and music experts, aiming to improve the recommendation power of Spotify's tools. The company has hundreds of music editors and experts around the world. A Spotify spokesperson said the generative AI tool allows human experts to “extend their innate knowledge in ways that were never possible before.”
Data for a particular song or artist captures certain characteristics: certain musical features, and the song or artist it has been paired with is typically among the millions of listening sessions whose data the AI algorithm has access to. Gathering information about the song is a fairly easy process, including year of release, genre and mood – from happy to danceable or somber. Various musical features, such as tempo, key, and instruments, are also identified. Combining this data linked to millions of listening sessions with the preferences of other users helps generate new recommendations, and makes possible the leap from aggregated data to individual listener assumptions.
In its simplest form, “Users who liked Y also liked Z. We know you like Y, so you might like Z,” is how AI finds matches. Spotify says it works. “Since launching DJ, we've found that when DJ listeners hear commentary alongside personalized music recommendations, they're more willing to try something new (or listen to a song they might have otherwise skipped),” the spokesperson said.
If it works, it won't be just the listeners who get relief from the pain point. This great discovery tool is also useful for artists looking to build relationships with new fans.
Everyone is trying to figure out how to balance familiarity and novelty in a meaningful way, says Julie Knipe, founder and CEO of Music Tomorrow — which aims to help artists connect with more listeners by understanding how algorithms work and how to work with them better. And everyone relies on AI algorithms to help make this possible. Whether she says the balance between discovering new music and staying with established styles is a central unresolved issue for all involved, from Spotify to listeners and artists.
“Any AI is only good at what you ask it to do,” Neby said. “These recommendation systems have been around for over a decade and have gotten pretty good at predicting what you'll like. What they can't do is know what's on your mind, specifically when you want to venture out into a new musical niche or category.”
Spotify's Daylist is an attempt to use generative AI to take into account established tastes, but also diverse contexts that can shape and reshape listeners' tastes throughout the day, delivering new recommendations to suit different moods, activities and emotions. It's possible that such improvements will continue, and that AI will get better at finding a formula for how fresh a listener wants it to be, Nebbi says, but she added: “The assumption that people want to discover new music all the time is not true.”
Most people still return, more or less happily, to familiar musical terrain and listening patterns.
“You have different profiles of listeners, curators, experts… People place different demands on AI,” Nebbi said. “It's hard to surprise experts, but they're not the majority of listeners, who tend to be more casual,” and whose Spotify use, she says, often creates a “comfortable backdrop” to everyday life.
Technology optimists often talk about an age of “abundance.” With 100 million songs available, but many listeners preferring the same 100 songs a million times over, it's easy to understand why they're searching for a new balance. But Ben Ratliff, the music critic and author of Every Song Ever: Twenty Ways to Listen in the Age of Music Abundance, says algorithms are less a solution to this problem than more entrenching it.
“Spotify is good at catching on to popular sensibilities and creating a soundtrack to them,” Ratliff said. “For example, its Sadgirl Starter Pack playlist has a great name and nearly a million and a half likes. Unfortunately, under the banner of the gift, SSP simplifies the peripheral complexity of depression in young adults into a small, relatable package” The yearning music works, makes the cliché Difficult music and sensitivity are formed more quickly.
Organizing work done by real people with actual preferences remains Ratliff's favorite. Even a good playlist may have been made without much intention or conscience, he says, but it's just a developed sense of recognizing patterns, “whether they're obscure patterns or widely known patterns.”
Depending on the individual, AI may have equal chances of becoming either a utopian or a dystopian solution in a world of 100 million paths. Ratliff says most users should keep it simple on their music streaming journeys. “As long as you realize that the app will never know you the way you want to be known, and as long as you know what you're looking for, or have some good prompts ready, you can find a lot of great music on Spotify.”
