Will AI Destroy Music?!

                      Will AI Destroy Music?! By Ashton Hummler

You and I are both living through it right now. The rise of artificial intelligence. With the recent breakthrough of ChatGPT, people everywhere are concerned about robots taking over the world…! Or at least our jobs. As a music major, I thought that I would be safe from the AI takeover because I thought that it could never touch music. Sure, it may be able to do simple unskilled labor, but it could never mimic the creativity and soul that comes with the art of music. However, now AI can create music based off of user submitted prompts instantaneously, which is much more worrying for a musician such as myself! So I just had to know how advanced the AI was with its music creation capabilities, and if I truly should be worried about my future as an artist. 

So what can AI already do? There are plenty of programs that can create royalty free background music for youtube videos or other visual media. Services like Amper Music, AIVA, and Soundful can do this for you almost immediately. All it takes is feeding the AI a couple of prompts on what genre you’d like, what instruments you want, and what type of mood you’re going for. Then boom! You’ll have a decent sounding backing track for your video within a minute or two. Amadeus Code and Boomy are two services that assist in music production with machine learning. Amadeus Code is able to write melodies automatically. Boomy is able to use and manipulate AI generated loops to potentially arrange full instrumentals with the help of the user’s personal taste and guidance. MuseNet is a program that can make original compositions using references of other classical or modern composers. You can tell it to make compositions inspired from any known composer, from Bach to Duke Ellington. Fun fact, MuseNet is made by the same developers who brought us the wonderful: ChatGPT.

Speaking of ChatGPT, let’s throw it into the hot seat. Since ChatGPT is a text based program, it can’t make any sounds, but it can write lyrics! ChatGPT can write any song on any prompt you give it… and I mean any prompt. 

For example: Here is a song ChatGPT wrote about squirrels going to space! 

Verse 1:

In a forest far away,

Lived a group of squirrels one day,

They dreamed of adventure and more,

Yearning to explore.

Chorus:

Squirrels going to space,

Flying high in outer space,

Exploring worlds far and wide,

The universe their guide.

Verse 2:

They built a spaceship strong and true,

With nuts and acorns, their fuel too,

Blasted off to the sky,

Their hearts soaring high.

Chorus:

Squirrels going to space,

Flying high in outer space,

Exploring worlds far and wide,

The universe their guide.

Bridge:

They gazed at stars and planets,

Danced among cosmic comets,

Their tails flicking in delight,

A sight that was truly out of sight.

Chorus:

Squirrels going to space,

Flying high in outer space,

Exploring worlds far and wide,

The universe their guide.

Verse 3:

As they journeyed through the stars,

They met other creatures from afar,

Aliens of all shapes and sizes,

And made new friends, full of surprises.

Chorus:

Squirrels going to space,

Flying high in outer space,

Exploring worlds far and wide,

The universe their guide.

Outro:

The squirrels returned home one day,

Filled with stories and adventures to say,

But their hearts longed to roam,

And explore the great unknown.

Wasn’t that amazing!? I already feel inspired! I’m going to title it Space Squirrels! I could produce this song right now and it could potentially be amazing for the Alvin and the Chipmunks movie where they go to space. Obviously, I could’ve told ChatGPT to write about something more… normal. Heartbreak or opulence are very common things to write songs about, and ChatGPT can do that too. When I was testing out the songwriting capabilities of ChatGPT, I realized all the songs had the same structure as the first. That structure being: Verse, Chorus, Verse, Chorus, Bridge, Chorus, Verse, Outro. This can easily be avoided however by just telling ChatGPT to make longer verses, or to remove the bridge, or whatever structural idea you have in mind. You could have a song with 6 different choruses… I don’t know why you’d want that but you can! I also realized all the writing in the songs were very weak in terms of wordplay or figurative language. Of course, a robot couldn’t delicately play with language the way a human can! Or at least… that’s what I thought. I simply just told ChatGPT to write a song about heartbreak and to add more figurative language and wordplay, and it gave me this:

Chorus:

My heart is a broken record, playing our song,

The melody so beautiful, but the words so wrong,

I thought our love was a rose, blooming so sweet,

But it wilted away, leaving me incomplete.

This is  my favorite excerpt because I didn’t want to underscore the true artistic mastery that was Space Squirrels, but this is actually, what we in the business call, FIRE. The idea of the “broken record” playing the “song” of the heartbreak over and over again because of the speaker’s inability to move on is pretty powerful. The melody being beautiful but the words being wrong!? That sounds like such a profound yet simple way to express the contrasting and paradoxical emotions that can be associated with heartbreak. I was not expecting this from a heartless piece of code at all. The metaphor of the rose is a bit basic, I will admit. However, I still find it impressive that the AI was able to “set up” for the delivery of the second line. It showed the rose blooming to then destroy it in the next line. You could argue this is very bare minimum song writing, but it still shows the ability of ChatGPT to not only be able to use metaphors and wordplay, but for it to set them up as well. 

So how is AI being implemented into the industry itself? Currently, artists record labels are trying to get their hands on it to get ahead so they can pioneer the field and, of course, make the most money. So it shouldn’t come as a surprise when I say that Capitol records signed an AI rapper named FN Meka.

Here is what he looks like. He was claimed to be a rapper powered to be completely powered by AI, meaning his songs would be written and performed by an AI voice over AI composed music. So… I guess that’s it then. The need for human artists is over, the machines have already taken over…

Is what I would say if he wasn’t a total fraud!!

There’s a whole lot of controversy surrounding the “concept” of FN Meka. To start, he isn’t even an AI. He’s just a rendered out animated model, and his voice performances are from an artist named “Kyle the hooligan”; his vocals and music were never generated by AI at all. He was also dropped by his own label for being too controversial. Images and videos circulated of the FN Meka model in music video scenes where he was shown as a victim of police brutality. FN Meka also has several lyrics where it can be heard rapping the N-word. FN Meka is supposed to be a racially ambiguous virtual rapper. These aforementioned actions show the rapper perpetuating negative stereotypes about African Americans, so the label dropped him because of all the bad press. Besides, the design of FN Meka himself doesn’t exactly scream “racially ambiguous”. In the end, it turns out FN Meka was just one massive sham, which still brings us back to our main question.

Will AI destroy music?

Well… did electronic music destroy music? Did it even destroy certain artists or genres? Did electronic synthesizers eliminate the many instruments that came before it? Well…no. The emergence of electronic alternatives to music are just that… alternatives: not replacements. To quote Robert Harari, the head of our music department at Stevens, he said: “AI like other digital innovations of the past is a powerful tool that musicians can use to enhance their creative output further than they could without.”. Musicians that take the time to learn and master the new technology can push the boundaries of what music is, creating new and interesting art for the world to consume. They can also do this without undermining the many artists that have come before them. Acoustic instruments and live performances would never die out no matter how advanced technology gets… because people like them! All types of music have listeners, from classical to country, even if the “technology” behind some of these genres is simpler or “inferior”. That could simply be a stylistic choice, like how some artists will mimic the technological limitations of vinyl records in genres like Lo-fi hip hop. There will always be an audience for the meticulously handcrafted music of man despite how good artificial intelligence gets at imitating it. 

If you want to know what I think… I think us musicians will be just fine. At least I hope I do. I like to believe that artificial intelligence will just be a tool that will only empower the current and any future generations of music. I mean, I already use AI to remove any background noise from my recordings and to help me get the best mixes for my vocals. So it’s already being used by me as a tool to enhance my own creations. I do want to remain positive that AI music won’t debilitate us, but empower us. Maybe one day in the far future, AI will be able to make perfect sounding music that sounds identical to a man made song. However, I still have hope that we musicians will never die out, and we’ll always be pushing the boundaries and always innovating the ever evolving field that is music. 

Work Cited: 

ChatGPT. Accessed on 5/3/2023. URL: https://openai.com/blog/chatgpt . 

Conteh, Mankaprr. “So What the Hell Was Fn Meka, Anyway?” Rolling Stone, Rolling Stone, 31 Aug. 2022, https://www.rollingstone.com/music/music-features/fn-meka-controversy-ai-1234585293/.