Creativity & AI: Process Over Product

A few weeks ago I saw a post on LinkedIn that brought up a topic that I hadn’t considered before. The post was written by Jack Diserens and focused on retention and pride when using AI music composition tools like Udio and Suno. The crux of the article was that because it is so easy to “compose” music utilizing one of these tools - no investment or effort - they become one trick ponies. Because the user invests almost no effort in the process of composing, they have no pride in the product. Because of this, users tend to play with these tools a few times and then move on when they get bored of them. And THAT is the dirty little secret of these AI music tools. Just like the company I run, MusicFirst, the success of these businesses rely on strong retention rates - meaning people keep coming back to them and perhaps more importantly, keep subscribing and paying for them. Without strong retention numbers, investors in these companies will start to invest elsewhere. All of this made me think of just how important the process is when it comes to musical composition. Have you ever had that moment when you write something you really like and can’t wait to play it for others. It is a special moment made even more meaningful by the fact that you composed it. AI tools simply don’t afford you that same experience, and I believe that is a good thing - for both composers and students. Without any expending ANY effort in the process there is simply no pride in the product.

My homebrewing setup

For those of you who know me, you know that I have been an avid home brewer for nearly 30 years. Over that period of time I have progressed from brewing in my kitchen with “Duncan Hines”-style recipe kits, to a professional brewing set up. I have brewed hundreds of batches of beer. When folks find out that I brew, they inevitably ask me the same question “What is your favorite style of beer to brew?” The answer is simple: whatever I’m currently brewing. I absolutely LOVE the process of brewing. The funny thing is that I rarely drink more than a few glasses of what I brew. I either give it away to friends and neighbors, or I end up pouring it down the drain. For me, the fun is all in the process of brewing. Everything from cleaning, to formulating a recipe, to the actual brewing process (which takes about 3-4 hours). Over the next few weeks there are a few other steps - from dry-hopping to transferring to a secondary fermenter to kegging or bottling. I love every step. And most importantly, I LOVE when folks try my beer and enjoy it. While I love that moment, the process is what its all about for me.

I think that the same can be said for ANY creative endeavor. Process Over Product is a prominent theme in Montessori Schools and the world of art, and it is a central component of my own creative efforts. Composing music is process-oriented. For me, I like to sit down at a piano and improvise for a while when I’m composing. Inevitably I will play something that I find interesting or even like. It could be a bass line, a melody, or a chord. When I hear something I like I focus in on it and try to expand on it after experimenting with what might come next. If I like what the direction of the musical idea, I might take out my computer and open up a notation program or digital audio workstation so that I can record the idea. Sometimes I continue the process in the software until I’m either finished with the piece or I run out of inspiration. I often look at composing like solving a musical puzzle with some emotional input and problem solving skills. Sometimes I solve the puzzle and feel a sense of satisfaction, and sometimes the puzzle goes unsolved - to either be picked up later or abandoned altogether. If I do end up abandoning it, there is no regret because the process makes it worth the investment of my time.

While crafting prompts into a AI-powered music composition tool circumvents almost ALL of the creative process, the lack of the investment of time and personal connection to the actual music means that you may have little to no satisfaction in the results. If you type in a bunch of prompts and you get something that sounds pretty good, what did you actually do? You typed some prompts. You did not compose music - the AI algorithm did all of the work. When GarageBand first came out, I clearly remember music teachers and composers warning that using loops “wasn’t composing” because you were using pre-made audio or MIDI clips. In my opinion, creating something in GarageBand takes WAY more investment than typing in “Write a song about pumpkin spice lattés”. It’s easy to create something that sounds terrible in GarageBand, and it takes effort to compose something that sounds good. Regardless of whether its “good” or “terrible”, when you press the play button you are playing something that you stitched together yourself. When I taught my middle school students how to compose, it took WEEKS for them to come up with something that they were happy with. That process WAS the learning. The end products that they produced (usually using GarageBand by the way) weren’t necessarily the most important thing. It was teaching them the process and then having them go through that process that made it a meaningful learning experience. Even the students who didn’t identify as musicians gained a lot from the experience. One of the best moments of teaching for me was watching students hear something they wrote with a look of astonishment and then calling over their friends to come and listen. THAT is what the process is all about. Sure it’s interesting and sometimes breathtaking to hear what an AI algorithm composes based on the prompts that you entered, but aside from being impressed or even scared with the underlying technology, I don’t think any student would claim that they actually wrote the song. Further, without their personal investment in the composition process, they may not have any reason to be proud of the results.

Each day my opinion of AI music composition tools becomes more and more complicated. On the one hand, I think that the technology that powers AI music composition tools is both remarkable and scary. And on the other, I find it really troubling that they have been trained on countless copyrighted musical works, and they don’t credit those works in their results. There are also many more ethical and philosophical concerns that run through my mind as well. Should you explore them on your own? With your students? YES. Use the results as inspiration for your own creative efforts. Grab a chord progression, a snippet of a melody or lyric that you like, or a drum beat. As Jack Diserens recommends in his LinkedIn post: Once an AI song is generated, allow users to tinker with it. Transform it like a ball of clay. Add their fingerprints to it. This will drive ownership -> Ownership drives pride -> Pride drives retention and sharing.

What do you think? I’d love to know how you regard the process vs. the product. How do your students view it? Let me know!