An unhappy listener, possibly suffering from the effects of prolonged exposure to AI audio. Photo Credit: Faustina Okeke
Just 3% of adults can tell the difference between machine-made music and the real thing – at least according to a worrying study from Deezer, which says it’s now being inundated with 50,000 AI tracks every day.
The service tapped Ipsos to spearhead the underlying survey last month. And according to the newly released results, 9,000 adults, hailing from the US, Canada, Brazil, the UK, France, the Netherlands, Germany, and Japan, participated.
Per the resource, those participants were asked off the bat “to listen to three tracks and determine whether or not they were fully AI-generated.” 97% failed to answer correctly – with 71% admitting to being surprised by the results and 52% indicating that they felt uncomfortable about misidentifying the audio.
(The involved companies opted against including the tracks’ duration and genre particulars, besides the actual titles, in the report.)
Unfortunately, the far-from-ideal findings don’t end there. 45% of “music streaming users” – referring to about 6,800 persons as opposed to all respondents – “would like to filter out 100% AI-generated music from” streaming, the study shows.
As a result, 55% of users wouldn’t like to filter out AI slop, a concerning stat because 70% nevertheless agreed that “100% AI-generated music threatens the livelihood of current and future musicians/artists/composers.”
Meanwhile, 40% of streaming-user respondents said they’d “skip without listening to 100% AI-music if they came across it.” Additionally, 80% agreed that audio made entirely by AI should be labeled as such. To date, however, Deezer is the only on-demand platform identifying and labeling machine outputs, which currently account for over one-third of the music arriving on the service daily.
Specifically, Deezer “is receiving over 50,000 fully AI-generated tracks every day – accounting for more than 34% of the total daily delivery,” according to the report.
As previously explored by DMN, try as they might, robots cannot replicate human authenticity; in general, AI audio is hard on the ears. (Quick, which artists do you think are likeliest to lose undiscerning listeners to robots? Spoiler alert: Probably not quality-minded indies.)
But from a straight numbers perspective, that doesn’t mean it won’t drown out new music and rising talent in a sea of sound. AI slop is pouring onto real artists’ profiles without authorization, infiltrating fans’ recommendations, diluting the royalty pool, and securing spots on popular playlists.
(On Deezer, AI tracks “are automatically removed from algorithmic recommendations and are not included in editorial playlists,” according to the platform.)
Furthermore, it’s all but impossible to overlook the unprecedented trend’s rapid acceleration. In January, Deezer pointed to 10,000 daily AI-track uploads – a figure that jumped to 20,000 in April and then 30,000 in September en route to cracking 50,000 today.
