According to Mashable, a new study from French music streaming service Deezer and research firm Ipsos reveals that 97% of people surveyed couldn’t tell when a song was fully AI-generated rather than human-made. The majority of those people—52%—felt uncomfortable about their inability to detect the difference. Despite this, 66% said they’d try AI music at least once out of curiosity, and 46% thought AI could help with music discovery. However, strong opposition emerges around transparency, with 80% wanting AI music clearly labeled and 70% believing AI threatens artists’ livelihoods. The survey also found that around 50,000 fully AI-generated tracks now upload to streaming services daily, accounting for 34% of all daily music releases.
The transparency problem
Here’s the thing that really stands out in this data: people aren’t necessarily against AI music in principle, but they absolutely hate feeling deceived. When you’ve got 80% of listeners demanding clear labels and 72% wanting to know when platforms are recommending AI content, that’s a pretty clear message. It’s the classic “don’t trick me” response that we’ve seen across so many technology adoption curves.
And honestly, can you blame them? If you’re spending your time and emotional energy connecting with what you think is human art, only to discover it’s algorithmically generated, that feels… violating somehow. The survey numbers show this discomfort translates directly into action—45% would filter out AI music entirely, and 40% would skip it if they encountered it.
Why can’t we tell?
So why are 97% of people failing this musical Turing test? Part of it’s obviously that AI has gotten scarily good at mimicking musical patterns. But there’s another factor the study hints at: a lot of popular music is already pretty formulaic. When you’ve got genres built around predictable chord progressions and repetitive structures, it becomes much easier for algorithms to replicate the sound.
Think about it—if you’re trying to generate classical jazz improvisation or complex progressive rock, that’s one thing. But four-chart pop songs with simple melodies? Basically AI’s sweet spot. The technology is advancing in the spaces where the human creativity bar was already… let’s say accessible.
The artist concern
The survey reveals something really important about how people view the creative process. While they’re curious about AI, they’re deeply protective of human artists. 65% don’t think AI should train on copyrighted music, and 70% see AI as threatening artists’ livelihoods. Only 11% think AI music deserves equal chart treatment.
This creates a fascinating tension for streaming platforms. They’re getting flooded with AI content—50,000 tracks daily is absolutely massive—but their users clearly want human creators prioritized. How do you balance discovery algorithms that might surface compelling AI content against user preferences for human-made art? It’s not just a technical challenge—it’s an ethical and business one too.
Where this is headed
We’re seeing the exact same pattern play out that we witnessed with AI art and writing. Initial curiosity gives way to demands for transparency and ethical boundaries. Platforms like Pinterest have already responded by letting users filter out AI content completely. Music services will likely face similar pressure.
The big question is whether streaming services will treat this as a labeling problem or a curation challenge. Do they just slap “AI-GENERATED” tags on everything and call it a day? Or do they develop more sophisticated ways to separate human-led creative processes from fully automated ones? Given that even tools like autotune require human expertise, there’s a whole spectrum of AI assistance versus AI generation that needs addressing.
One thing’s clear from this Deezer and Ipsos research: listeners want to make informed choices about what they’re consuming. The platforms that respect that desire while still enabling discovery and innovation will probably come out ahead. The ones that treat AI content as just another track in the library? They might find themselves dealing with some very unhappy subscribers.
