Is your music smarter than you are? The technological imperative and music.

Will Artificial Intelligence end new music and new artists?

As a PhD student at MIT I took courses in the history and philosophy of science and the structure scientific research.  A cardinal rule embedded in all of those courses was the “technological imperative”:  new technologies are inevitable and will be developed regardless of impact.  Shortened,  it says, “if it is possible,  we will design it and build it, like it or not.”

As an MIT alum I now receive “Technology Review” magazine, chronicling new technologies developed by MIT and other tech center graduates, and summaries of reports from the Media Lab and various departments.  In 2019 I read a report from the Media Lab entitled Towards New Musics: What the future holds for sound creativity byTod Machover and Charles Holbrow , reporting on progress toward generating music with AI and other technologies. The article explored AI as a way to enable a song to continue to grow and change after it was recorded and released.  An interesting concept, and one as a music writer I knew was  fraught with legal and financial consequences, so I began to keep an eye on “technology music”.

Last year Input magazine ran a article that popped up on my filters, “Spotify could soon replace real artists with AI music”  by James Fergus, that noted two patents by Spotify designed to put AI to work to catch plagiarized songs,  but when taken together could be the foundation of a system to use AI to create music that Spotify owns.  The company could post its AI music, competing with the artists it already stresses with miniscule payments and its global monopoly. Now,  I was very interested. This could be serious.

But the idea that music streaming services could actually use AI to eliminate the middleman/woman in the creation of music was just speculation.

It is no longer speculation, at least the part  about using AI to produce music that  may compete with real live songwriters and performers. My friend Shelly Peiken, the platinum songwriter, recording artist , and amazing writer,  used her February 18  ‘Confessions of a Serial Songwriter” blog to alert music lovers to the AI-based  technology, Jukebox. She  described Jukebox  as  “a system that intakes the genre and mood of a given song, the artist and snippets of lyrics, and then is given the task of ‘making’ a song that sounds like that artist could have written and performed it themself. Jukebox can mimic voices, instruments, scores, styles, symphonies, you name it. And it’s super vibey.”

As Peiken says, “this isn’t artificial intelligence. It’s artificial art”.  I say it is perilous, at least to songwriters and musicians.

Jukebox was developed by Open AI, a Elon-Musk founded research organization well-known to anyone from MIT as a cutting edge seedbed to move the technological imperative along, including in the direction of private profit.  Its investors include Microsoft, Reid Hoffman’s charitable foundation, and Khosla Ventures,  who are presumed to want to see some kind of financial or social return on their investment.

So how would Jukebox, or similar systems that are now on the market,  produce an ROI for investors?  One answer may be  to look at  Spotify.  If the hunch is correct that Spotify is exploring using AI to cut out the middlewoman and man in music by producing songs that compete directly with human-created music, this will revolutionize music – again – and not well for either the artists or the fans.  

We don’t know if Spotify, or any other streaming service,  has plans to insert AI- generated songs into its libraries, playlists, and recommendations.  But you can already listen to a Barbican Center-produced AI playlist on Spotify.   Since AI music costs little to nothing to create after the initial investment in equipment and data, large streaming services will have a strong incentive to, at the very least, further, cut royalty payments to humans , set up divisions that stream nothing but robot tunes. or just stop streaming human-made music.  

Peiken points  out the myriad legal and intellectual property questions that robot music raises, but for me, a music critic who tries to find and support rising talent, the bigger problem is that  AI could lead to the end of new artists, new music, new songs – real songs, not robot songs. 

Why? Because if the streaming services opt for free robot music that they own, that leaves artists with little income except live performances, occasional TV and film and commercial placements, and merch tables. For big stars with millions of fans that may not be a problem, but for rising or even mid-list artists, it will make a hard  living  even harder.

Publishing royalties may take up some of the slack, but I doubt it. AI companies like Jukebox could sell music much cheaper without mechanical or other royalties.  Why pay full price for a real song when you can get one just as good (or maybe better “designed” to sell in a particular market) for much less from an AI company?

Of course, songwriters have relationships with singers and labels and other people in the music food chain, so there won’t be a sharp end of music as we know it.  Just as the advent of streaming took a few years for Spotify and Apple to end the album as we know it and kill off the CD, AI’s impact won’t be immediate or total.  Popular artists will survive on their fan bases and live shows.  It is the struggling artists just finding their voices  and those passionately entering the music world with exciting new sound who  find that AI has cut them off from finding an audience.  We will have music, but the human side will fade and the robot side will rise. It’s the technological imperative.

But when it comes to creativity there is a huma imperative.  It’s not too late to educate, legislate, and participate to insure that humans remain in control of music and the human imperative, not the technological imperative, determines what we create,  listen to  and enjoy.

Patrick O’Heffernan, Host, Music Sin Fronteras radio

If you want to learn more about the future of music, here are some good places to start.

Shelly Peikin, Confessions of a Serial Songwriter,  shellypeiken.com/confessions (subscribe!)

J. Fergus, “Spotify could soon replace real artists with AI music,” Input, 12.7.2020. https://bit.ly/3sBoPEj

Kyle Wiggers, VentureBeat , “AI system infers music from silent videos of musicians”, MIT Computer Science and Artificial Intelligence Lab, Aug. 3, 2020, https://bit.ly/2ZUwbGz

Tod Machover and Charles Holbrow “Towards New Musics: What the future holds for sound creativity”, MIT Media Lab, July 26, 2019. https://bit.ly/2ZYcBsM

.