New research explores why people like AI music in private, but won't publicly replay it, explore the artist or share it.
University of Dayton marketing assistant professor Andrew Edelblum and his undergraduate co-author Joshua Poe conducted three studies to find:
- 88% of consumers say they prefer human-made music in the abstract.
- Within 60 seconds of actually listening, that preference disappears. People rate the same AI-generated song just as highly as one labeled "human-made."
- Even when listeners enjoy AI music, they remain 15–25% less willing to endorse it.
"We went into this expecting people to reject AI music outright, and instead we found something more interesting. They like it, they just don't want to be seen liking it," Poe said. "Being able to dig into a question that big as an undergraduate, and to do it alongside a mentor like Dr. Edelblum, is the kind of opportunity that shaped how I think about research and about my own career."
The pair also found the pattern appears specific to music. When they ran a parallel test using poetry — a creative form often compared to song lyrics — the effect did not appear, suggesting music may occupy a unique space among creative outputs because of the immediacy of the listening experience and the emotional responses it triggers.
The findings come as AI-generated tracks rack up millions of streams and fictional AI artists land major-label deals, raising new questions for streaming platforms about how and whether to disclose AI involvement to subscribers and to the human artists on their rosters.
"Working on this project with Josh has been a wonderful experience," Edelblum said. "We literally received our acceptance from Psychology & Marketing mere days before he graduated.”