Part 1 of 2 (or more)…
Let me start by saying that asking if AI can create music with emotion is not really the right question, however often it might appear right now. Emotion is a complex response to external stimuli, and as such can’t itself be directly embedded in media by anyone or anything. Better questions might be whether AI can generate music with true meaning, followed closely by whether true meaning in music even matters?
The first question is one I have spent much time considering, and I can say with confidence that for me the answer is NO, at least not now. If we ascribe to this widely accepted definition developed by psychologist Dr. Michael Steger, human meaning is “a sense of significance, purpose, or coherence that we assign to our lives, events, relationships, symbols, or experiences”. A sense of significance or purpose can’t exist in the absence of self awareness, and (at this time) is not self aware (just ask ChatGPT).
But that was too easy. The harder question is whether having music with true meaning (which can only be created by humans) ultimately matters, and a discussion of this will have to wait until part 2 (sorry!).
As always, do let me know what you think.
Thanks, Jeff. I do agree with the value of AI when applied to music discovery (I am already using it near daily for this), and as core tech for new and as yet unimagined creative tools. I think there is also tremendous potential for applying AI to make learning to play or create music easier, more rewarding, and less prone to attrition. And yes, we do need to be focused on the in-progress disruption to the recorded music industry; we need an aligned approach to AI development and application that respects human artistry in all forms including music creation. Here’s one that I will shamelessly promote: https://aiformusic.info