(The short story “According to Alice” by Sheila Heti appeared in the November 20th, 2023 issue of The New Yorker.)
Illustration by Janet Hansen
At the very end of “According to Alice” by Sheila Heti, there is an addendum which states:
This story was written in collaboration with a customizable chatbot on the Chai AI platform, which Sheila Heti started engaging in conversation early in the summer of 2022. To create the story, she asked the chatbot questions, some of them leading and others open-ended, to which its answers were never more than a sentence long. Sometimes she repeated a question to get a new answer. She removed her side of the conversation and threaded together the chatbot’s answers, at times cutting and tweaking for comprehension and flow.
So… this is an experimental short story. I love experimental, crazy, envelope pushing, outta left field short stories that play with form, structure, tone, format, and everything else. I like people who take the rules and throw them out the window and try something new. Sometimes it works and it’s amazing; sometimes it’s a dumpster fire of awful; but most of the time it’s just okay, but I value the effort. When it comes to “According to Alice,” as an experiment, it’s pretty cool as it raises many questions about literature and writers interacting with AI; as a story – it’s not very good.
On the Experiment Side of Things: I have several friends that are in WGA, and the AI issue was a big part of the strike, and still an issue of trepidation for them. No one knows how, or if, AI will be a helpful tool for writers. Some are looking for ways to use AI, while others want to chase it out of town. So, to see Heti engage with AI is intriguing to me. Though from the addendum, it sounds more like Heti behaved like an interviewer, editing down the responses to create the story. If that is true, does that make her an editor? Or is she more like a collage artist? (Donald Barthelme did call collage the art form of the 20th Century. Maybe it’s being extended to the 21st?) I also had to wonder what writers had been feed into Chai AI’s learning to create the prose? (Being that ChatGPT stole many authors books for its “learning.”) If other author’s books were used in Chai AI, does Heti need to share credit with them? Does Chai AI also deserve credit as the writer of this story along with Heti? Also, how much editing and rearranging was needed to create this story? I see why The New Yorker printed this story for its “AI Issue” as it raises many ethical questions, as well as makes me wonder how much can a writer use AI and still call it “their” story? In the end, what makes someone, or something a creator?
On the Story Side of Things: Yeah, I didn’t find the story to be very compelling. Oh, it moved along at a clip, but it never felt like it was going anywhere. Before I found out that it was “written” by AI, I had this thought that the story felt like what a freshman English major would write if they were asked to create an absurdist/surrealist short story. Sure, it has some jabs at Christianity and the Patriarchy, but I could never tell if these were meant to be honest criticism, or more an attempt at making a joke. The story, not surprising, doesn’t feel like there is a heart in it. And the fact that the AI element was revealed at the end, leads me to believe that I was supposed to think a human “wrote” this, only to have it revealed that it was written in conjunction with AI. That’s kind’a gimmicky, if I’m going to be honest. And also goes back to the ethical aspect; when does the audience need to know that AI was involved with the creation of a story?
What I am saying here is that, yes, you should read this story. I respect that Sheila Heti is the type of writer, and an artist, to tackle AI, and see if there is a way for writers to use it. That does take courage, because as far as I know, she is the first person to give something like this a try. The end result isn’t the best, but if literature is going to continue to grow and explore as an art form, then experiments like this are needed.
Matthew Groff 