by Andrew Lee

CS 470: Music and AI

As I mentioned in my last critical response, one motivating factor for me to take this class was because of the rise of AI art and the anti-AI art movement that began last year. At the time, I felt so conflicted as both a CS student studying AI and a music composer. After the deep discussions and innovative assignments that I've had in these past 10 weeks, I now want to respond back to my feelings back then. Therefore, this is a message to Andrew Lee from 2022.

Dear me,

“Is it even right to do this?” we asked ourselves as we saw posts of AI art on our Instagram feed and articles on the anti-AI art movement. Frankly, we was terrified. We were fascinated by all the cool AI systems people have made so far, but this led us to seriously reconsider how we feel about AI. So we decided to take this class. As a programmer and a music composer, what would this mean for us? Well, now that I took the class, let me tell you what I think.

To start off, I'm still scared. You're not going to believe this, but Google developed this prompt-based music generator called musicLM just last month. What does this mean for us as a composer? I don't know. I'm hesitant because of how things are going in the art community. But ultimately, I'm hopeful that society won't be heartless enough to just replace human composers with artificial ones. I believe that people will always find the works of humans much more valuable than those created by an AI. You can give me a beginner-level romantic piece and a Chopin-level one generated by an AI, and I'd still prefer the former because there was human intention, effort, emotion, and thought put into it. I can't speak for the rest of society, but I'm sure a good chunk of people think that too.

You must be wondering what it's like to use AI in music creation without it replacing the compositional process. From all the things I made in this class, my answer to you is "with interaction." One thing I made was an audio mosaic tool. Basically, it's a song generator that takes in some microphone input, and plays the most similar sound from the sound database you give it. With this, I made a synthesizer that plays snippets of music from "The Great Showman" based on what I play on the piano, and performed a duet with the synthesizer. How cool is that!? There's actually so much potential in developing new ways to make music with an AI. It doesn't replace you, but instead expands on the things you can do, if used with the right intention of course.

The reason we're so fearful of AI is because the industry only keeps making more AI that replaces things that we, humans, can already do: self-driving cars, AlphaGo, art tools, educators, etc. Why make machines that can do what we can do? Please make machines that help us do more things, or assist us to do our jobs better. Not machines that replace us entirely. If we can make a future like that, maybe AI won't be that scary and we can still put our trust into people.

Ultimately, don't be resistant to AI because you're scared of it. Face it, and make it better. Let's build technology that we can truly appreciate and feel comfortable with.

See you in a few months,

Me from 2023