Humans No Longer Able To Distinguish AI-Generated Music From Human-Made Tracks: Survey
The global music industry is undergoing a significant shift. Recently, Ipsos surveyed Deezer users and found that AI-generated music is so realistic that most listeners are fooled. Even music professionals and hardcore fans who have listened to a massive number of tracks cannot distinguish between AI-generated and human-generated music.
In this survey, over 97% of people couldn’t tell that the music tracks are AI-generated, not human. Deezer CEO Jeronimo Folgueiro stated that “AI is revolutionizing the music industry and we need to ensure it evolves in a way that benefits artists, fans and the entire ecosystem.”
He added, “We conducted this survey to better understand how people perceive AI-generated music and to guide our efforts in building a transparent and fair music streaming environment.”
Survey details: How the test worked

The survey was a collective effort by the global research firm and the music streaming platform. Over 9000 people from 8 different countries took part in the survey.
Citizens of the US, UK, Germany, Brazil, Japan, the Netherlands, France, and Canada were the core participants.
The participants listened to 3 short music clips. Out of these, two were created by AI, while one was human-made. The participants were asked to pick one they felt was human-built. However, only 3% got it right, while the remaining failed.
| Category | Key Findings | Population Base |
| General Awareness of AI | 98% have heard of AI | General Population (n=9000) |
| 72% have used AI at least a few times | General Population | |
| 55% feel curiosity as a primary sentiment toward AI | General Population | |
| 19% feel trust as a primary sentiment toward AI | General Population | |
| AI in Music Creation | 46% believe AI can help them discover more music they like | General Population |
| 51% believe AI will play a significant role in music creation in the next 10 years | General Population | |
| 51% believe AI will lead to more low-quality, generic music on streaming platforms | General Population | |
| 64% believe AI could reduce creativity in music production | General Population | |
| Recognition & Consumption | 97% couldn’t distinguish AI-generated music from human-made in a blind test | General Population |
| 52% felt uncomfortable not being able to tell the difference | General Population | |
| 66% would listen to 100% AI-generated music at least once, out of curiosity | Music Streaming Users (n=6791) | |
| 45% would like to filter out 100% AI-generated music from their platform | Music Streaming Users | |
| 40% would skip 100% AI-generated music without listening | Music Streaming Users | |
| Transparency Expectations | 80% agree AI-generated music should be clearly labeled | General Population |
| 73% want to know if their streaming service is recommending AI music | Music Streaming Users | |
| 52% say AI music should not be included in the main music charts | General Population | |
| Only 11% believe AI music should be treated equally on charts | General Population | |
| 58% believe their platform never recommended AI music | Music Streaming Users | |
| 25% are uncertain whether AI music has been recommended | Music Streaming Users | |
| Fairness Toward Artists | 65% oppose using copyrighted material to train AI music models | General Population |
| 70% believe AI music threatens the livelihood of current and future musicians | General Population | |
| 73% say it’s unethical to use copyrighted music without the artist’s approval | General Population | |
| 69% believe payouts for AI-generated music should be lower than for human-made music | General Population |
Real Example: “Walk My Walk” By Breaking Trust
It was a famous country song. It hit number one on the US Billboard Hot 100 in 2025. This entire song was generated by AI, including the vocals and lyrics. Until this survey happened, many fans were unaware of this reality.
It was shocking news for many participants, and some called it “too perfect,” while others praised its emotional power. However, others raised questions regarding AI transparency, artist rights, and listeners’ emotions. They said there should be clear labeling on all AI-generated tracks.
The most common AI tools like Udio and Suno use “diffusion models to create music. They make it from scratch just like humans do. These models are actually trained on millions of human songs, tracks, and sounds. However, unlike humans, AI didn’t create music step by step. It’s just all of a sudden a formula that produces genuine music in just seconds.
