Skip to main content

Uses of AI in Academia

by Addison Eldin
26 March 2024

Over the past few years, I have incorporated AI and text generation technology into the undergraduate courses I teach. At the most basic level, I have added an AI policy to my syllabi, an idea encouraged by Pitt's Writing Institute. The resultant conversations, far from frustrating, have set the semester off on the right foot. Last term, when explaining the syllabus policy on the first day of class, a student asked how I would even determine whether someone used text generation technology inappropriately. I explained some indicators, but was clear that I would not be using online evaluation tools for AI use. The anxiety about potential false accusations of academic dishonesty was alleviated, and instead, we had a conversation about ethical uses of ChatGPT and other software in the writing process, with appropriate accreditation.

I have also incorporated the use of AI into the content of my courses. In recent semesters, I have been teaching Narrative and Technology. It’s a writing-intensive English literature course that is taken by many students in Digital Narrative and Interactive Design, a major that integrates the English department and the School of Computing and Information. While the course is available to all undergraduates, it tends to attract students who are interested in the relationship between digital technology and the decidedly human terrain of storytelling.

Roughly halfway through the term, in Narrative and Technology we move from media like print short stories, comic books, film, and television to electronic literature and games. We consider some unique or prominent dimensions to digital storytelling, such as the increased role of interactivity and the new kinds of narrative it enables. The stronger the presence of the technology in generating elements of the narrative—and the less apparent the hand of a specific author—the less clear how our familiar ways of understanding narrative can be applied. 

Studying the role of technology in generating elements of the narrative leads into a session focused on the intersection of narrative and text generation AI such as ChatGPT, an edge case of interactivity and the lack of a clear human author to the text. Students engage with AI Dungeon prior to class; they choose a genre and character role, which ChatGPT turns into a scenario. They then type in what their character does or says, or even type in direct narration, and the text generation technology produces new narrative beats in response.  

The goal of using AI Dungeon is not just to enjoy the novelty of an authorless story—if we require authors to be human. It’s to ask whether the probabilistic, beat-by-beat digital “author” paired with a human author produces something we consider to be narrative. Is this an authored representation of a story, fictional or not, akin to previous media? Is the lack of a clear structure in AI Dungeon stories, without a clear ending and arc to the narrative that reaches toward that ending, any different from a serial narrative where the author doesn’t have any end-goal in mind? 

Admittedly, not every course is focused on, well, narrative and technology. But I think this activity is worth noting because it shows how the quickly proliferating applications of text generation AI can serve as the basis for engaging students and considering pressing questions in the relevant field or discipline. Narrative theory has long had to consider the effects of digital technology on storytelling, and here, students can apply their understanding developed in class to a cutting-edge conversation in the field with unclear ramifications. 

If you are interested in seeing more about how I have approached incorporating text generation technology into my teaching, please visit my website, or see either of my publications in TextGenEd, “Genre Generators”  and “Narrative and Text Generation AI.” The latter provides a more detailed breakdown of the activity discussed above, and was reworked for this newsletter in keeping with the publication’s Creative Commons license: Creative Commons Attribution-Noncommercial (CC-BY-NC).