What is the impact of artificial intelligence on streaming platforms? Is it used only as a tool to finish songs or even to produce them?
According to Deezer's managers, Spotify and Apple Music competitors born in France and owned for more than a third of Access Industries (Warner Music Group, Dazn), every day about 10 thousand songs generated by artificial intelligence are loaded on the platform, equal to about 10% of all music published daily on Deezer.
The estimate was made using a software that sifted the uploads. To counteract the phenomenon, Deezer intends to mark the content generated by the AI, including the tracks that use Deep Fake items, also excluding them from the recommended playlists to users.
For Alexis Lanternier, CEO of Deezer, “artificial intelligence upsets the musical ecosystem every day, producing an increasing amount of content that floods platforms such as ours”. Ai can potentially have “a positive impact on the creation and consumption of music, its use must however be responsible and oriented towards the protection of the rights and revenues of artists and authors”.
It is no small theme. Just a few years ago, when the music generated automatically or almost it was of infimal quality, the problem of songs produced with the AI seemed irrelevant. With the refinement of artificial intelligence tools, it is no longer and is indeed becoming increasingly important especially in relation to the streaming platforms and playlists that provide for a passive listening of music by users, for example those designed to encourage or evoke a certain state of mind or accompany daily activities.
In the future, the contents generated by AI could in fact remove space and therefore revenues to the musicians. And let's not talk about a distant future. The Confédération Internationale Des Sociétés d'Uteurs et compositeurs, an international non -governmental organization that works to protect the rights and interests of the authors, estimates that in three years the AI could subtract almost 25% of artists' revenues. There is talk of around 4 billion euros.
Streaming platforms are not said to have an economic incentive to stem the phenomenon in a decisive way. The journalist Liz Pell, author of the book Mood Machine: The Rise of Spotify and the Costs of the Perfect Playlistexplains that “the story of the so-called fake artists has been running from 2015 or 2016. I had taken for granted that they were do-it-yourself cheats that tried to screw the system. The stories of single scammers of streaming services have never particularly interested me. Focusing on the individual cheats has always seemed to me a distraction from the systemic deception that is underway. Instead, discover that the so -called fake artists actually part of a strategy conceived within Spotify, where there is a small team that interfaces with these specific licensee, and that this program has even a name has surprised me a lot ».
Among the many musicians who took the field against the use of AI there is Paul McCartney who in an interview granted to the BBC said he was worried about the English bill that would allow technological companies to train artificial intelligence models using Copyright protected works, except for explicit opposition from the artists. It is another crucial aspect of the question: before generating songs with “sense”, the AI models need to store a large amount of music. Last year Sony, Universal and Warner sued Suno and Udio, platforms that allow you to create songs with a simple written command: they would have been trained by violating copyright, or giving them passage songs without the necessary authorization of those entitled. The English law would make this operation easier.
“To collect,” said McCartney, worried about the younger artists whose works are less and less protected and remedied, “should be the creator of the song, not a big tech”.