My research paper will be about the AI interpretation of spontaneity/improvisation in jazz when generating music. I provide examples as to why AI does not have a sense of spontaneity or improvisation when it generates music. I will reference my firsthand experience with improvisation and compare/contrast AI’s. I want to understand the generative shape, form, and structure, and deduce if this is present in the music at all. I will research what else there is to AI’s generations besides mimicking of human creation, finding anything that has been tweaked with the code that allows it to step closer towards seeming more human. I will reach into other forms of art that involves spontaneity/improvisation like poetry and painting.
To provide context for my argument, I will use J. L. Moreno’s Theory of Spontaneity-Creativity from Sociometry and the Science of Man to familiarize myself and the reader about where this comes from and the psychology behind why it is ultimately human.
I will then move into applying my knowledge and transcribing what the AI is playing and compare/contrast it to what a human would play over the same exact chords. Then, I will explain why the AI reached this output by demonstrating a reverse engineering approach, stripping down each layer of the generative tool and how they all work together to create this music, referencing Mycka, Jan, and Jacek Mańdziuk’s “Artificial Intelligence in Music: Recent Trends and Challenges.”
I will finally argue my point that AI does not have any sense of spontaneity and it never will, referencing Cole, Ross. “The Problem with AI Music: Song and Cyborg Creativity in the Digital Age.”, and possibly even wokring with OU’s jazz director.
I will end this paper with why AI is ultimately a negative thing for the music industry, referencing Wilf’s The Inspiration Machine Computational Creativity in Poetry and Jazz
I also anticipate working with Miles Davis’s Miles: The Autobiography and method books for learning improvisation (Charlie Parker Omnibook), but have not found a specific use case yet.
Is the ending going to be on why AI is ultimately a bad thing for music, or specifically a bad thing for music genres that need improvisation (like jazz)? I would be interested in the former but it seems like it would be harder to argue if most of the paper is dedicated specifically to how AI can’t improvise/be spontaneous. Perhaps somebody could say that improvisation isn’t necessary for more ambient genres or something like that, therefore AI can’t be bad for music as a whole.