Reading Response #9: "Humans in the Loop" + "Experimental Creative Writing"

Alanna Sun

November 21, 2021

CS 476A, Stanford University

Reading Response: What Counts as Art

I found Allison Parrish’s talk in “Experimental Creative Writing with Vectorized Words” really interesting, serving as an example of the AI-systems with “humans in the loop” that are argued for in the article by Ge Wang. It was so strange seeing words and their meanings mapped to mathematical values. The computationally generated poems that Parrish read sound so beautiful and convincingly human. Because the poems are generated in sequences of similar-sounding words, reading them aloud captures the physical motion that your mouth goes through to form these sounds, and hearing them reiterates this kind of circularity and fluidity of the poem that is allowed by the use of all the homophone-adjacents. However, I personally would have found the poems more powerful if Parrish had written them herself after using the algorithm to generate the words that she would use. For me, the human significance of art is captured by the intention that is put into what is presented. I think it’s really cool that the algorithm is capable of generating something so beautiful, but upon learning that the poem is, in fact, generated by an algorithm, I lost the motivation to read into the poem and form the connections that I otherwise might if I thought that the poet was a human. This is because I feel like poetry is meant to communicate a way of thinking, and part of the joy of reading it is trying to understand the poet’s perspective. As a result, knowing that there is nothing deeper behind the poem kills a bit of the joy of reading it. This goes back to the larger question of what counts as art, and what are the intangibles that are lost when we automate art-making. In this case, I appreciate that Parrish’s use and design of her AI-system are very much intentional, artful, and creative. Yet, the fact that the words themselves and the order of the words did not come from her still leaves me feeling like something’s missing. I can’t quite put my finger on it, but I wonder if others feel the same. At the same time, one could argue that the reader is the one who gives life to a poem and imbues it with meaning. After all, once a poem is out in the world, it no longer belongs to the poet, and even outlives the poet (or at least the good poems do). If a computationally generated poem can make me feel deeply, then perhaps it shouldn’t matter that it’s derived from an algorithm. But does it only make me feel that way because I’m assuming that its poet is another human being, someone I can feel a connection with? In that case, would the poet be lying to the reader, because it/she never corrects this assumption?