A post on Instagram claims that research shows that 97% of listeners can’t distinguish between AI and human-made music.

When we investigated this claim, we traced the statistic to a survey commissioned by the music streaming platform Deezer and conducted by the market research firm Ipsos, released in November 2025. The study aimed to quantify just how sophisticated generative audio has become and involved a substantial sample size of 9,000 respondents across 8 countries.

In the experiment, participants were subjected to a “blind listening” test where they heard a mix of tracks composed by human artists and those generated entirely by AI algorithms. The results were stark: 97% of the respondents failed to correctly identify which songs were AI-generated in a blind test with two fully AI-generated songs and one real song. This indicates that for the vast majority of casual listeners, the gap in quality and “human touch” between synthetic and organic music has effectively closed.

Beyond the headline statistic, the report highlighted a significant emotional response to the use of AI technology in music. Approximately 52% of participants reported feeling “uncomfortable” upon realizing they couldn’t tell the difference, while 71% expressed genuine surprise at the capability of the AI.

Deezer has used these findings to underscore a critical issue in the streaming industry: transparency. With the platform reporting over 50,000 AI-generated tracks uploaded daily, they argue that these survey results validate the urgent need for “Artist-Centric” payment models and clear labeling tools, ensuring that listeners know exactly what—or who—they are listening to.

The claim is accurate regarding the specific findings of the Deezer/Ipsos study. While this represents a single dataset from a company with a vested interest in AI detection, the high sample size suggests a genuine shift in the audio landscape: human ears can no longer reliably distinguish artificial from organic music.