
Spotify CEO Gustav Söderström just dropped a bombshell at SXSW in Austin: the days of fighting with your "bad" music recommendations are almost over.
In a move that brings ChatGPT-style control to your playlists, Spotify is rolling out a massive update to its Taste Profile. For the first time, you won’t just wonder why the algorithm thinks you love sea shanties — you’ll be able to tell it exactly where it’s wrong.
How the new "Taste Profile" works

Using a built-in Large Language Model (LLM), Spotify is giving users the power to "mold" their recommendations using natural language. According to a recent Spotify blog post, the experience is designed to be conversational:
“If something doesn’t feel quite right, you can tell us by flagging when your profile misses the mark—asking for more or less of a certain vibe, or simply sharing what you’re in the mood for.”
In the coming weeks, users will finally have the chance to actually review their curated profile and alter their recommendations to their preferences.
What you can change
This isn't just a "thumbs up/down" button. You’ll get an inside look at how Spotify describes your taste in music, podcasts, and audiobooks, then use prompts to tweak it:
Boost specific artists: "Give me more Sabrina Carpenter, less Doja Cat."
Refine genres: "Switch my focus from J-Pop to K-Pop this week."
Control the "vibe": Dial back the energy for work or crank it up for the gym.
The feature begins its rollout in the coming weeks, starting exclusively with Premium listeners in New Zealand. After this initial testing phase, the AI-powered Taste Profile is expected to expand to global markets.
Bottom line
Spotify’s Taste Profile is improving to the point where users are gaining more control over what the app advises them to listen to next. Before, they were only able to exclude certain playlists and even individual tracks from affecting what pops up through their recommendations.
With this upcoming update, Spotify users will have more access to their Taste Profile so they can see how it decides what they should listen to next and tell it exactly what they’re looking for.
