The contents of Chapter 5 integrate the most meaningful and grateful things I have learned at CCRMA over the past year. Many of the principles in Chapter 5 have been continuously mentioned by Prof. Ge Wang in [Music 356: Music and AI] and [Music 128: Stanford Laptop Orchestra]. Revisiting and reviewing those principles made all the memories of my first year go through my mind like a flash.

As written at the bottom of p229, these principles easily extend beyond computer music. Especially in my case, these principles provided me with a new perspective on AI music. Before coming to CCRMA, most of my research experience was about applying neural network-based AI to musical tasks, but in the research process, I never thought about how people would actually use such AI or how they would interact with it. However, what I realized during my first year at CCRMA is that the artistic quality and playful/joyful nature of an AI tool as a musical medium ultimately depend on how we design interaction between human and AI, and what the interface for such interaction should be.

At the Audio Engineering Society (AES) 2023 conference, two key notes, Hank Shocklee and Charles Alexander, the pioneers who established 'Hip-hop sound' had a opening talk to celebrate the 50th anniversary of Hip-hop. What they did was just telling us about their story of meeting new analog or digital audio equipments (sampler, sequencer, turtable, etc) during the audio revolution in the 80s and 90s, how they just played with it and explored new sounds. While they were talking that, they were really excited, joyful, and enthusiastic.

While I was listening to that key note talk, I wondered why such a thing is not happening with AI music tool. Why AI music tools are not playful? Why it does not feel like a toy that encourages interaction and enjoyment? I think, for now, all the existing AI music tools are like something outside of us which does not extend us. In order to AI music tools obtain artistic quality and playfulness, there needs to be more structural discourse on the human-AI interacion design and interface design so that we can feel them like an extension of ourselves.

Principle 5.19. Interfaces should extends us (and not replace us). We want tools, not oracles.

+ As someone who has used hemis the most recently due to a series of performances, I witness hemis being broken and broken every day, and looking at historical photos of such hemis being born was very interesting. It's like looking at a photo of a friend I've known for a long time when he/she was a newborn.