Two 2026 studies report the same perception gap: in blind tests, listeners rate AI-generated songs as good as or better than human ones, but the moment they learn a track is AI, their interest in replaying it — and their willingness to pay — falls.
2026-08-18 · by Moe Ameen
Two separate 2026 studies landed on the same finding: people tend to like AI-generated music on its own terms, and stop liking it once they know it was made by AI. When listeners judge tracks blind — by sound alone — AI-generated songs score as well as or better than human-made ones on quality, enjoyment, and the desire to hear them again. The gap opens only after the AI origin is disclosed, when relisten interest and willingness to pay drop. The change is in the label, not the audio.
The first study, by Jana Friedrichsen and Julia Schwarz of Kiel University and Michel Clement of the University of Hamburg Business School, was summarized in a ProMarket piece on May 4, 2026. Across three empirical studies (the experimental ones run in 2024), the researchers varied whether listeners were told a pop or electronic-dance track was AI-generated. Told nothing, listeners actually rated the AI songs' likelihood-of-relistening higher than the human ones; told the track was AI, that desire to relisten and their willingness to pay both decreased — an effect driven mainly by pop listeners. The second, by University of Dayton marketing assistant professor Andrew Edelblum and co-author Joshua Poe (accepted at the journal Psychology & Marketing in mid-2026), found that 88% of consumers say in the abstract that they prefer human-made music, but that stated preference vanishes within about 60 seconds of actually listening — people then rate an AI track the same as one labeled human-made. Even while enjoying it, listeners were roughly 15–25% less willing to publicly endorse the AI song. The effect was specific to music; a parallel test with poetry showed no equivalent penalty. As Poe put it: "They like it, they just don't want to be seen liking it."
Both studies sit against a broader souring of sentiment. Luminate's "Generative AI in Entertainment 2026" report found overall interest in AI music sliding from roughly -13% to -20% over 2025, with the sharpest negativity among Gen Z and Gen Alpha listeners and toward tracks that mimic existing artists. Keep the framing honest: these are survey and experiment samples in specific genres, so the exact percentages are directional, not a verdict on all AI music. The durable, cross-study signal is the perception gap itself — disclosure changes how a song is received without changing a note of it.
The immediate move is to publish the research while it's still being argued about. Feed the findings into [Kompozy](/) and one source becomes a blog explainer of the disclosure penalty, a carousel contrasting "blind-test rating" against "rating after disclosure," quote cards of the 88% and 15–25% figures, and a short captioned video — fanned out across an owned blog plus the eight primary social platforms and email in a day on [Autopilot](/glossary/autopilot), behind a per-post review gate. A clear, timely explainer of a counterintuitive study is exactly the kind of content that earns reach while the topic is hot, and it positions you as the person who read the paper instead of the headline.
The deeper read is the one that shapes how you use any AI tool, Kompozy included. The research says the discount isn't on the work — it's on the anonymous, unattributed AI feeling behind it. That is precisely the gap a consistent brand identity closes. To be clear about what Kompozy is: it generates video, images, carousels, blogs, and newsletters (and layers music into formats like Marketing Shorts) — it is not an AI song generator, and it won't pretend to be. What it does own is the trust layer the studies point to. A single [Persona Brief](/glossary/persona-brief) governs voice across every output, Gemini face-lock keeps a recognizable persona consistent shot to shot, and HyperFrames renders pixel-exact brand styling — so your posts read as *you*, not as generic AI filler, which is the exact difference between the blind-test score and the post-disclosure one. And because norms are moving toward mandatory labeling, the honest play is to disclose plainly and let a strong, familiar identity do the reassuring — see our coverage of [the EU AI Act's labeling rules](/news/eu-ai-act-content-labeling-rules-2026) and [Spotify's AI-persona policy](/news/spotify-ai-persona-labeling-policy). Kompozy's job is to make the AI-assisted output feel authored and on-brand across every platform, so the label costs you as little as possible.
In blind tests, yes — when listeners judge songs by sound alone, they rate AI-generated tracks as good as or better than human-made ones on quality, enjoyment, and desire to hear them again. One study even found listeners were more likely to want to replay the AI songs. The preference reverses only after they're told a track was made by AI.
Two 2026 studies converged on it. One was by Jana Friedrichsen and Julia Schwarz of Kiel University with Michel Clement of the University of Hamburg Business School, summarized in ProMarket on May 4, 2026. The other was by University of Dayton marketing assistant professor Andrew Edelblum and co-author Joshua Poe, accepted at the journal Psychology & Marketing in mid-2026.
The evidence points to a social effect rather than a sound one. Listeners still enjoy the track privately, but they become less willing to endorse it publicly and less willing to pay for it. As one researcher summarized the pattern: people like it, they just don't want to be seen liking it.
Not automatically. The disclosure penalty showed up for music but not for a parallel test with poetry, so part of it is a music-specific cultural attachment. The transferable lesson for creators is narrower: authenticity and a trusted, recognizable identity matter more than raw output quality when AI origin is visible.