Reading Response
I absolutely loved this reading and video lecture. To me, this is one of the most interesting topics in the field and represents one of my primary research areas. In fact, reading the article in the Human-Centered AI blog felt uncanny — I wrote about many of these same ideas in my application to Stanford! Similarly, with Allison Parrish’s fantastic presentation, I felt a mix of excitement and jealousy since I have been dreaming of doing exactly the kind of experimental linguistic art she demos, and have been working on projects on the side with lyric and poem generation (even using the same tools like CMUDict) except less flexible and less efficient. So, on the one hand I’m annoyed because I had independently had many of the thoughts and project ideas I read/watched, but at the same time it makes me incredibly excited and motivated to continue along this path. I hope we work on and discuss these kind of topics in “Music and AI” next quarter! I feel a sense of affirmation in my values, and confirmation that I am in exactly the right place to keep pursuing my research interests. To me, I’ve had a love-hate relationship with Artificial Intelligence for many of the same reasons Ge outlined in his article. I loathe the idea that Music + AI is or should be just directed towards building “computer musicians” whose job is to mimic, emulate, or replace human music. I totally agree that this kind of framework is appropriate and needed in other industries or disciplines — economics, statistics, etc. Even there, I also feel uneasy with the attitude that AI is this godlike oracle whose ability and function to predict and reproduce is treated as the end-all-be-all solution to everything. However, I feel like the tools AI offers simply cannot be ignored — the power of machine learning algorithms and the pace that AI is evolving can and should be harnessed in creative fields. But only with a more human-centered way, with a more careful treatment of their role in the creative process. Out of a human-computer collaborative system can emerge art that is truly unique, and something that cannot be accomplished with either tool alone. And, I’d posit, not “better” or “worse” in “quality” to purely human-generated music, but simply a categorically different medium. A medium whose content can, just like “normal” music, be diverse in genre, popularity, emotion, and so on. Perhaps an “AI musician” will be a worthwhile and genuine role to fill after all, just not in the way people often initially imagine.