“Discover Weakly?”: Perceptions of Spotify's Algorithmically Generated Music Recommendations
Year: 2025
Type: dissertation
Abstract: Spotify, a large music streaming service used across the world, features AI-generated playlists, or collections of songs that dynamically adjust to user listening behaviors. Academics and journalists have raised concerns that Artificial Intelligence (AI) and AIgenerated playlists reduce listening diversity and monopolize the means with which people discover new music. While past research uses quantitative means to examine user behavior, fewer studies address user perceptions of the role of AI in music discovery. To address this gap, I conducted an interview study with 13 participants and analyzed the results to elucidate 1) how they utilize three AI-generated playlist features: Discover Weekly, Daylist, and DJ 2) what factors impact which feature they choose to use and when, and 3) whether their musical avidity (or engagement with music) impacts their perceptions of the expanding role of AI in music listening. Key findings include: 1) Participants regard the Discover Weekly playlist as the most influential to their new music discovery; 2) participants enjoy the blending of new and familiar music into themes from the Daylist, and 3) participants desired the most customization with the DJ feature.
DOI: 10.17615/cya7-t614
Author: Hartsock, Jeremiah
Institution: University of North Carolina at Chapel Hill
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7082149798
Cites:
Cited by:
Related to: