A.I. Is The Future of Music

Unfortunately, Artificial Intelligence is going to be the future of music.  The result is going to be like the Napster debacle times 1000, because it’s going to effectively end the music industry.  It might even end music as a hobby.  It will definitely end music as a commercial venture, and by proxy will destroy the idea of the “rockstar”, or any kind of fame, notoriety, or prestige associated with being a musician.  In my humble opinion, it’s going to reduce the role of a musician to whatever a cryptographer is now.  There will be exceedingly few people who do it “professionally”, and there will be the millions of people who do the musical equivalent of crossword puzzles in the Sunday paper.   

What is this upcoming apocalypse, you might ask?  The answer, quite simply, is computer-created original music.  And by computer created, I mean artificial intelligence that creates new music at the push of a button.  If this seems far-fetched, it’s already here.

It works by feeding an algorithm a bunch of similar music.  Then, after programming in the appropriate music theory rules and constraints, the A.I. is able to compose completely new songs and albums in seconds.  Currently, the results are mostly genre specific, or even cases where a single artist was used as the “input”.  The “rules” of music theory are mostly universal, and when you apply those with similar “rules” that dictate genre and playing style, music becomes something that is surprisingly formulaic.  As a life-long musician and composer, I can attest to this.  Music is really not hard to create once you know the rules, and there are no reason why these rules, and the subsequent choices that they offer, cannot be managed by a machine. 

  You can experience original Death Metal music created by a neural-network here.  While this is subjective, you can make the argument that this computer-composed music is indiscernible from music created by humans.  I would be hard pressed to tell this apart from a human band, and I’m sure those totally unfamiliar with Death Metal would be unable to discern them.  And this isn’t going to stop with Death Metal- because these types of algorithms are possible for any genre, you can be sure other types of music are just around the corner.  

The real question arises once neural-networks that create original music become commonplace enough to use commercially, which is bound to happen anytime now.  Think about it form a music-licensing perspective;  Let’s say you are making a movie/video game/fill-in-the-blank and need to license original music to use in your content.  Are you going to pay the standard rate of several thousand dollars to acquire this music from humans, or are you going to open an app, hit the space bar, and use the music that your $100 software created?  Think about it.  Licensing music from humans entails paying publishers, paying songwriters, etc., while the music producers themselves scramble to recoup costs for musicians, studio time, etc.  It can be a complex ordeal with many people involved.  Creating music using A.I, on the other hand, will literally be as simple as choosing a few parameters, pressing “go”, and enjoying the royalty-free    Most movie, TV, and video game music directors I know are working with ever-shrinking budgets, so I can’t see how they wouldn’t chose to do this once it becomes an option. 

It’s also probably fair to say that “electronic” music has become increasingly popular over the past decades, and this is often in opposition to music that is made on traditional instruments like guitars and acoustic drums.  It is also worth mentioning that while this has probably made the public more receptive to music that’s not made by humans, A.I.-created music is not limited to electronic music.  So ask yourself, what are you going to do when all the music you hear in your day-today life is created by computers instead of humans?