Article written by John Wilcox, Communications Instructor
My students are working on This I Believe essays, which are short statements of personal belief told through a personal story. This activity occurred after topic brainstorming and some early planning of their essay, but before the actual essay was written.
The purpose of this activity is to show students how valuable their own voices and experiences are in their writing.
I asked my students to use a large language model to generate an essay on their This I Believe topic using a prompt like “I am a college student in a first-year writing class. Generate a 350 word This I Believe essay on [TOPIC.]”
After the students generated the first essay, I asked them to get the AI to revise the essay by telling it to add personal details to specific paragraphs, for example “in the second paragraph, tell a story about how my dad taught me [TOPIC]. (I, of course, encouraged my students to use made-up details if they wanted.)
After the second essay was generated, I asked my students to get the AI to revise one last time to add even more detail.
When my students had an AI generated essay on their topics with some of their own specific details, I asked them to post the essay to a discussion board and answer the following questions:
1.) This essay you generated never existed before and wouldn’t have existed without your input. Do you feel ownership of the essay? Do you feel any pride in the results the AI generated based on your prompts?
2.) What can you do better than the AI?
We then talked in class about their responses. I’ve only done this exercise so far in one class, but here are the results:
17 students said they did not feel any ownership of the essay and no pride in the results.
1 student felt pride that they were able to get the AI to generate something pretty close to what they were thinking and feeling.
Of the 17 students who didn’t feel any connection to the AI generated essay, independently 10 of them reported some variation of feeling “creepy” or “weirded out” by reading the generated essay (the discussion board was set to hide replies until the student posted, so they weren’t just agreeing with each other.)
All my students who actively participated in the in-class conversation talked about how the AI “writes better than them” but can’t tell their story in a convincing way. I argued that means the AI cannot “write better than them.”
The largest part of the conversation centered on the inexplainable eeriness and creepiness so many of them felt. There was a real sense in the room of some kind of “violation”—that the AI took their ideas and experiences and spit something out that resembled what they wanted to say but wasn’t it. One student said (seriously) it was like her shadow wrote the essay.
AI can write convincingly “human” writing. Especially paid versions of LLMs are getting pretty scary-good at mimicking the writing voice of a typical college freshman. It is getting harder and harder to convince students not to turn to AI to do their mental labor for them. My hope is that by assigning things students want to write and showing them firsthand that AI can do a lot but it can’t live their lives for them, fewer of them will turn to AI. And honestly, I hope the ones that do turn to AI in their stress or apathy at least go through the extra mental effort of trying to get the AI to generate something closer to what they actually want to say.
0 Comments.