How Mozart Wrote 759 Trillion Songs
AI-Generated Art, from Mozart to Mhi'ya Iman Le'Paige
According to Google’s AI, Mozart wrote over 600 pieces of music. Depending on how you define “wrote”, the real number is more like 759 trillion.
Have you ever played one of those “build your own drag name” games, where you pick a first and last name from two different columns depending on some random variable like your birthday or the first letter of your last name?
According to this one, I’m Dame Ivana Bang Bang!
The Musical Dice Game
There is a similar game attributed to Mozart, like a “build your own waltz”. It’s called Musikalisches Würfelspiel, or Musical Dice Game.
The game works like this: you roll a pair of dice, add up the numbers you get, and then you pick a measure of music from a list which corresponds to the number you rolled.
Here are the numbers I got by rolling a pair of dice 16 times, adding together the sums for each roll: 10, 7, 4, 6, 3, 12, 7, 2, 12, 9, 7, 6, 7, 8, 3, and 2.
So you pick the 10th measure in column A, the 7th from column B, and so on.
Mozart already did the hard part of creating the database of music you have to pick from. So you just look up the corresponding measures and you can string together your own little song!
I used this website to piece together this song based on my dice rolls, and you can play around with it to make your own!
If the dice fall differently, you’ll get a slightly different tune. The number of possible ways you can roll a pair of dice 16 times is 11^16, however the 8th roll always gives you the same measure, and on the 16th roll there are only two possible measures.
So the total number of possible songs this game can produce is 2*11^14, or 759,499,667,166,482.
Who owns all these songs?
Various different composers in the 18th century made their own versions of the musical dice game, and it became a popular way for anyone to become a composer. Since the number of possible songs was so huge, it was likely that every piece of music you created with the game was unique, and never heard before.
When I first learned about this musical dice game and the sheer number of possible songs it could produce, it got me curious.
Is it accurate to claim that Mozart wrote all of these 759 trillion songs? Sure, he composed each measure, but he couldn’t have listened to all the possible combinations of them. Am I a composer just because I rolled some dice and looked up numbers in a column? Of course not — I don’t have any musical knowledge, and I didn’t put any thought or creativity into those notes. It was all random.
But randomness can be artistic too. One of the most famous American composers is John Cage, who is best known for composing a song made of 4 minutes and 33 seconds of complete silence. He made music out of randomness too — like stones falling off an ice sculpture as it melts, or a toaster falling over.
“Everything we do is music." - John Cage
Generative art
Generative art refers to art that’s created by an algorithm, and usually includes some element of random chance. Instead of a human being making the creative decisions, an algorithm makes them, whether it’s through random dice rolls, the first initial of your last name, or a computer program.
I’m sure you’ve all seen AI-generated art on your timelines by now.
On season 16, we watched Mhi'ya Iman Le'Paige walk down the runway in an outfit inspired by an AI-generated image she saw online which was in turn inspired by the prompt “What if Rupaul’s Drag Race was made in the 80’s?”
Image generators like DALL-E and Midjourney can create images out of a single sentence.
Just as the musical dice game produced a song taken from a randomized collage of Mozart’s music, AI-image generators produce images inspired by a randomized collage of billions of example images.
The result is an entirely new set of pixels that no one has ever seen before, but no one can agree on who owns it.
Artists have raised concerns about copyrights and intellectual property when it comes to their artwork being used in training datasets for these AI image generators. I think those concerns are reasonable and important. There should be a healthy, ethical relationship between art and technology.
I have to admit I find AI fascinating. I’ve started taking some online courses about machine learning out of curiosity. Eventually I’ll do a post on here and on TikTok about the math behind these algorithms.
Humans are incredible, whether we’re creating artwork, or creating algorithms that create artwork.