Music 256A Reading Response 9

Elena Stalnaker | Fall 2021

Reading Response ~ "Humans in the Loop: The Design of Interactive AI Systems" by Ge Wang and "Experimental Creative Writing With Vectorized Words" by Allison Parrish

I think Parrish's work is an excellent example of what we might actually want from AI. She uses AI word-vectors as a tool for creating art. The AI reads to me as somewhere between an instrument to be used and a partner worthy of respect, both of which seem like healthy human-robot relationships. This relationship between poet and machine seems to change depending on the way in which Parrish is using the word-vector data for a given poem. "Frankenstein Genesis," Parrish's program that mixes the two paragraphs from "Frankenstein" and the Bible, seems like a tool to be played and experimented with, to amusing effect. However, Parrish's program "Articulations," a "random walk through phonetic similarity space," harnesses the power of her phonetic word to vector map to procure entire lines of phonetically similar poetry. Something about the machine procuring entire lines of poetry, instead of single words, makes it feel more like a collaborating artist than a mindless tool. Sure, it makes choices differently than a human artist would given a similar task, and with a slightly less mysterious set of criteria. But sampling lines of poetry from other's works based on phonetic association sounds like something a human poet or lyricist might do, drawing on their memory and experience to create something new. At the same time, I suppose that even if the action looks very human on the surface, the program's reason for doing it is very different. Even though this poetry generating software may seem reminiscent of sci-fi-esque broad AI, it's doing what it's doing because Parrish gave it very specific instructions, while a person might engage in a similar composition exercise because they were moved by some emotion, or just by playful curiosity (is curiosity an emotion?).