François Pachet has the capacity to change the music industry as we know it. You probably don’t know who he is. I don’t expect you to. He’s a French scientist and foremost expert on music composed by artificial intelligence. Yes, music composed by computers. Why should you care? Francois was hired in July by Spotify to little fanfare. Previous to Spotify, Francois served as the Director of Sony’s Computer Science Laboratory in Paris, which focuses on developing “flow machine” technology that can collaborate with musicians and compose its own music as well. You read that correctly. AI that can create its own music with absolutely no human element involved.
In an opaque blog post, Spotify explained that Pachet “will focus on making tools to help artists in their creative process” which barely scratches the surface of his talents, experience, capabilities. Pachet founded the Computer Science Laboratory in 1997 and has made computer music, machine learning, and machine improvisation his life’s work. He has 15 patents, has been published in over 80 academic journals, and is an accomplished composer in his own right. He’s considered to be an AI music pioneer. Pachet’s hire at Spotify raises serious questions about the music industry’s future and Spotify’s increasing control over the artists, labels, and distributors that depend on it.
You see where this is going. Spotify hires the leading AI expert in the world under the guise that they want to enhance an artist’s ability to compose music. Imagine a world where Spotify slowly phases out artists and labels and replaces them with machines that create music specifically designed for playlists. It’s already happening. The average consumer of music couldn’t tell you who sings their favorite song on Rap Caviar or New Music Friday. However, they can tell you what playlists they follow religiously. Spotify and Apple Music have conditioned us to listen to playlists versus individual artists. The artist doesn’t matter anymore. It’s not by accident that Spotify enables you to listen by genre and mood. It’s by careful design.
Further, consider the economics of replacing artists and labels with artificial intelligence. Currently, it’s projected that Spotify pays out approximately 50% (down from 70%) of its revenue to artists and labels. Spotify’s margins are rumored to be meager despite their consistent growth. Spotify posts a loss every year. Plus, they’re said to be preparing a direct listing on the NYSE in the coming months. Spotify is financially incentivized to push artists and labels out and replace them with AI. They’d be instantly profitable, their stock price would shoot through the roof, they’d leave Apple Music in the dust, and Amazon would swoop in and acquire them making Daniel Ek a very rich man. The writing's on the wall whether labels or artists want to read it or not.
In fact, it’s already happening. Sony’s CSL Research Lab released the first ever AI composed song called “Daddy’s Car” in September of 2016 financed by an ERC grant which is a project Pachet heads. The AI learned a variety of music styles from an extensive database of songs that allowed it to create a single that sounded similar to The Beatles. Don’t believe me? Take a listen below.
It’s not just Sony that’s experimenting with music composed by AI. Google is on the action as well with a project called “Magenta” that launched quietly in June of 2016. Magenta uses “machine learning to create compelling art and music”. In February of this year Spotify launched “A.I. Duet” that uses neural networks to allow you to play a duet with the computer in real time. You play a note. The computer plays one back.
What would prevent Spotify from pushing labels and artists out and replacing them with Sony and Google AI that can compose music? Complex global licensing deals that allow Spotify to offer the vast catalogues of music we love to stream for the low low price of $9.99 a month. Spotify depends on labels to keep these licensing deals intact, otherwise those labels could pull their catalogues in protest and Spotify would lose valued subscribers to the smaller (but growing) Apple Music. Spotify can’t afford for that to happen and they know that, especially with a looming IPO in the near future. What’s likely is that Spotify slowly introduces computer generated music over time until there’s no turning back. It’s a slippery slope that will only become more complex.
Despite the points against, Pachet’s hire should be a wake up call for an industry that’s largely asleep at the wheel. Labels thought the advent of Napster was bad. Imagine a world where labels are competing with machines that do nothing but create music solely engineered for playlist consumption. We reached out to Pachet to give him the opportunity to respond to this article. Instead, he put us in touch with Spotify’s “PR” team who responded with the following:
"Just to be clear, Francois and his team will focus on making to to help artists in their creative process, not replace them."
Despite the denial, Pachet’s hire should tell you all you need to know about where Spotify and the music industry is headed. It’s time we all wake up.