Technology as Threat

Arful Design Chapter 2

I am responding to Ge Wang's Artful Design, Chapter 2, in particular the interview on "One Person, One World." Ge's defense of computer music and music technology innovation felt like reading my own conversation history from when I was working at a music AI start-up. When my work came up, reactions were split between intrigue and distrust. The latter group would ask questions like, "Don't you feel bad putting real composers out of work?" The short answer was no.

The long answer: While every render our software made was unique, it was not something unheard of. The way we got to the output was complex and definitely innovative, but musically, it existed in the same sphere as stock music libraries. There are countless examples of creative video scoring, but Academy Awards are not won using stock music offerings. It is an industry designed to deliver extremely specific emotional content to the media's end consumer. What does it even mean to innovate in a space where success necessitates tapping into a listener's pre-existing understanding of music?

Particularly for orchestral tracks, many stock music composers use sample packs rather than live recordings. What is the difference between a human programming an algorithm to put together samples and a human putting them together manually? When people refer to "real composers," all they mean is the latter, but algorithms are no less rooted in human authorship. Technical details aside, the process of my job was very similar to any other stock music composer's: thinking up chord progressions and rhythms, and making every instrument work as an ensemble. New technological interventions are always met with resistance. The argument that computers should stay out of creative processes makes me imagine someone at the onset of industrialization, arguing that butter must be churned by hand.

The only time I felt conflicted about my work at Amper was while developing trip hop. For copyright reasons, we couldn't sample real-world songs, so a workaround was created to simulate the sampling aesthetic quintessential for the genre. Sampling is such a human-to-human transfer, tying together artists across culture, time, and space. Mimicking that process algorithmically stripped the sounds of their usual layers of embedded meaning. For stock music applications, this doesn't matter to the final listener, but it was the first time I had any concerns about the broader implications of computerizing a creative process.

It seems like the goal of technological innovation for a while has been to convincingly recreate the physical world, but once something passes the Turing Test, its true identity still often matters to its users. Cleverbot was fun to talk to because you knew it was a computer - if it had just been some guy, I wouldn't have bothered! As another example, Hatsune Miku is a completely digital singer, but her fans and listenership have tangible impacts no different from a traditional pop star. Like Ocarina with ocarinas before it, she does not emulate a pop star: she is one, and not because she pretends to be a physical person. The more society considers digital creations just as real as their physical counterparts, the less they need to mask themselves as physical objects. Maybe that is the techno-utopia to strive for: a world in which the digital can flourish on its own terms.