Why Spotify AI more than music is secret to keeping listeners

Music streaming apps are pushing users into the age of artificial intelligence with a limited track record of success. But AI-based recommendation tools from Apple, Amazon and pure streaming company Spotify We continue to move forward as Spotify’s newest approach to the future of personal music discovery relies on AI prompting across multiple formats. Experts say these technology investments could be critical to Spotify’s ability to build a moat around its business as its key input, music, becomes commoditized on streaming apps.

A new ChatGPT integration recently released by Spotify allows users to connect their accounts directly to OpenAI’s generative AI chatbot. The launch is a welcome development for OpenAI in its broader effort to turn ChatGPT into a platform for third-party applications that run inside conversations. For Spotify, there’s a bet that personalized music and podcast recommendations will be enhanced through the now-familiar format of chatting with AI and letting it know what it wants.

Spotify users can request songs, artists, albums, playlists or podcast episodes by mood, genre or topic. The results appear in ChatGPT and open in the Spotify app for playback. Users can interact with the recommendation and provide specificity beyond what is possible with the classic “like/dislike” feedback option.

According to a Spotify spokesperson, these prompts are “an opportunity to uncover new tracks or revisit old favorites, or extend a ChatGPT conversation with music appropriate for the moment.”

Spotify said the integration is optional and users can disconnect at any time. It also said it would not share music or podcast content with OpenAI for educational purposes, addressing industry concerns about AI and copyrighted material.

Spotify also recently introduced Prompt Playlist on its music streaming app, a “live” feature that allows users to tap on an emotion or memory to create a custom mix.

Rival streaming services affiliated with big tech players are exploring similar AI features.

Apple is slowly layering AI into Apple Music. “Playlist Playground“The beta feature is closest to what Spotify is doing because it also focuses on chat-based AI interaction that allows users to make changes to recommendations via chat. Apple recently introduced AutoMixAnalyzing songs and automatically blending tracks by matching tempo and beats, eliminating silence between songs, adding transitions, etc. A technology that uses artificial intelligence for The company also introduced machine learning tools such as: Lyric translation and pronunciation features.

Amazon Music introduced a prompt-based playlist feature called Maestro Since mid-2024, it has allowed listeners to create playlists using text descriptions and even emojis. It remains in beta testing instead of the full version.

Spotify executives have repeatedly identified AI as central to the platform’s subscriber engagement strategy. One latest earnings callLeadership told investors that improvements in AI-driven discovery are central to keeping users engaged with the platform. “Our investments in personalization and artificial intelligence are paying off,” said Alex Norström, chief executive officer. “This means people spend more days and more moments with us in a month,” he said.

Spotify’s interactive iDJ feature, introduced in 2023, is used by nearly 90 million subscribers as of its latest earnings report, with users spending more than four billion hours on the app. Stimulated Playlists are “instantly available to power users,” Norström said.

“If iDJ is Spotify’s casual chat interface, Prompted Playlists are Spotify’s Deep Research mode,” he said. “It allows you to define and set the rules for your own personalized playlists; you literally write your own algorithm. … There’s nothing else like it.”

Music catalogs are commercialized, AI produces millions of songs

According to analysts who follow Spotify, executives’ hype about artificial intelligence may sooner or later become reality for the company. Although there are cases where musicians occasionally pull music from certain apps as negative headline issues arise — this It happened to Spotify due to its founder and former CEO Daniel Ek’s recent investments in defense technology; Competitors like Apple Music, Amazon Music, and YouTube Music offer largely overlapping catalogs and increasingly complex recommendation engines.

“The catalogs on Amazon, Apple and YouTube are similar to Spotify — nearly identical songs — just like Bing and Edge are nearly identical to Google,” said Michael Pachter, a senior advisor for digital media, sports and entertainment at Wedbush Securities who has covered the streaming industry as a research analyst for many years. (Wedbush never had an exclusive rating on Spotify shares.)

As Google’s search business faces its own AI threat, Pachter said this is also the best model Spotify will look at for how to maintain user advantage. “Google has managed to widen its moat by introducing a number of features that make the service even more difficult, including remembering my credit card and password information. I can’t even imagine switching from Google Search, and I think that’s what Spotify is trying to establish,” he said.

Switching costs may be small but significant. Users build libraries, create playlists, and train algorithms over the years. Each additional integration, whether with a car dashboard, a voice assistant, or now an AI chatbot (Spotify says it now connects to more than 2,000 device types), can further solidify the ecosystem.

“I expect this ChatGPT integration to be widely used and extremely successful by Spotify users,” Pachter said. “Others may try to do the same thing, but switching costs increase every time you put effort into creating your playlists on Spotify, and they rely on that,” he said.

Apple Music and other third-party apps offer tools to export playlists when subscribers want to switch music services.

Others on Wall Street weren’t as convinced as Pachter, but they were more bullish on the Spotify AI story from last quarter and less worried about the risks facing AI music creation tools disrupting platforms like his. Spotify’s stock price has fallen nearly 20% in the past year, but the stock has performed very strongly since its 2018 IPO.

“Spotify has addressed this concern head-on, arguing that AI bolsters rather than undermines its strategic position. By leaning into personalization, product innovation, and scale advantages, Spotify appears positioned to use AI to strengthen its platform, but speed of adoption and industry fit will remain key variables,” Bank of America’s research team, which rates the shares a buy, wrote in a February note after the latest earnings.

Spotify co-CEO Gustav Söderström said during these earnings calls that developing a music app that users can talk to and that fully understands each listener will transform listening from “a passive experience to an interactive experience.”

Mark Mulligan, managing director and analyst at MIDiA Research, a research firm that tracks the music market, says AI will be an integral part of music streaming behavior, but he is less convinced that the distinction between interactive and passive made by Spotify is the likely outcome.

“Streaming music consumption is split between passive and active,” Mulligan said. “But that doesn’t mean listeners are split in two; everyone is. Even the most active music listeners spend more than half their time passively listening.”

In fact, he says the broader trend is towards more passive consumption through curated playlists, as well as features like artist radio stations and AI DJs. “The direction of travel is toward more passive listening,” Mulligan said. Agency characteristics may represent a compromise, “a middle ground between passive and active listening,” he said. “It allows the user to expend a small amount of ‘lean forward’ effort in exchange for a large amount of ‘lean back’ listening.”

Typing a detailed command into ChatGPT might feel active, but Mulligan says, “The more the algorithm learns about the listener’s behavior and tastes, the better its recommendations become, and therefore the less the user has to lean in, thus moving the needle further toward passive listening.”

In this AI interface-first flow model, core content is important, but what ultimately makes the user feel rewarded is less important. For example, the ability to explicitly exclude artists or narrow them down by subgenre could enable AI-powered discovery to be more customized than traditional algorithmic playlists. If a listener enjoys 1980s rock bands like Bon Jovi and Guns N’ Roses but dislikes others from the same era, they’re easier to filter out. Spotify can normally predict that if you like A and most people who like A also like B, you’ll probably like B, but this doesn’t accurately represent how users’ tastes are expressed. “With GPT I can say ‘No Def Leppard’ and my listings will be deleted from them,” Pachter said.

Any predictions about the impact of AI on music, as in other industries, are guesswork. However, it is already clear that artificial intelligence has an impact on the idea of ​​a music catalogue. According to a recent Rothschild & Co Redburn report, text-to-music platforms like Suno reportedly produce around seven million songs per day; That’s roughly the equivalent of Spotify’s entire pre-AI catalog every two weeks. “This is a deluge,” said analyst Ed Vyvyan.

Söderström hinted that what is most important for the future is not the deep traces already in the stack, but the dataset yet to come. “We’re creating a data set that never existed,” he said on a recent earnings call. “We had a song-by-song dataset, but no one had a language-to-song dataset. … You might think of it as a canonical dataset, so is there a realistic answer to, say, what is workout music? There’s no realistic answer to what workout music is. … for the average American, it’s usually hip hop. For a European, it’s usually EDM. For many Scandinavians, it’s something like heavy metal or even death metal. Then again, most Americans, at least millions of them, are also death metal.”

“You can’t have a Masters commoditize this as a fact the same way you commoditize Wikipedia,” he said. “You actually need hundreds of millions of listeners in markets around the world who are constantly telling you what this means to a particular person.”