As Soon As (A)I Speak(s)
For Geneseo’s focus on AI in Ideas That Matter 2024, I created a two-workshop sequence to help students understand how a Large Language Model (LLM) composes poetry, and to give them the knowledge to compose their own poem for optional display in both electronic and physical form. The two workshops, titled “As Soon As (A)I Speak(s)” were held on September 20 and October 18, respectively. For anyone interested in understanding the goals and methodology of the workshops, or in cloning or forking them for their own teaching, I’ve deposited an instructor guide and my presentation slides in Geneseo’s institutional repository, KnightScholar. The slides are also embedded below.
Workshop One allows students to understand the basic operations of artificial intelligence and to briefly compare three different AIs, using them to generate three poems. Workshop Two focuses on the specific process of Prompt Engineering, in order to improve the poem the student is trying to write through developing critical reflection skills.
While a key goal of the workshops is to help participants understand social expectations for the art form of poetry and how poems construct meaning, the workshops could be adapted to run using a variety of other written modes and genres, such as fiction, science writing, song lyrics, and more.
Your post is a great resource for those of us interested in both AI and poetry (and the arts more broadly). I try to incorporate some amount of engagement with “art” in my classes that don’t already obviously touch on it, but I’ve stayed away from using poetry because frankly, I don’t understand it. Ok, that isn’t entirely true. More accurately: I don’t understand how to talk about in a product way with other people. In my own head and at my own pace I can pick a poem apart, chew on it, play around with connects and deeper meaning, and then eventually (a day a week?? a year???) settle on what I think the poem means. But I don’t know how to do this with others. Do you have any suggestions on how a poetry novice can do this?
Second, I follow the AI critic Gary Marcus on Substack and his most recent posting was about the hype around AI-generated poetry. He included a link to a paper by the computer scientist Ernest Davis titled “ ChatGPT’s Poetry is Incompetent and Banal: A Discussion of (Porter and Machery, 2024)”. Davis’ paper is responding to the mentioned study, which had the enticing title: “ AI-generated poetry is indistinguishable from human-written poetry and is rated more favorably.” I had to laugh as I read the takedown of the study, but the one point I wanted to raise to you is the idea that AI-generated poetry is more favorable *because* it is easier to understand. It affords surface level engagement with meaning and symbolism, using words that aren’t complicated to disentangle. How do we use AI in such away that it doesn’t water down the rich experience art affords us but which still allows us to find and make use of the benefits that AI produces?
Sorry if that turned into a ramble. lol
https://cs.nyu.edu/~davise/papers/GPT-Poetry.pdf