Deakin University

File(s) under permanent embargo

AI and the Future of Literary Studies

educational resource
posted on 2023-03-08, 02:17 authored by Andrew DeanAndrew Dean
The article has three main segments: 1) I begin by comparing two university mission statements, one generated by AI, one from the university at which I work. The counterintuitive outcomes of the comparison shows how powerful generative AI is at mimicking language that does not refer to anything. This language of make-believe is especially significant given the training data on which early natural language processing is based: the Enron emails. I contrast mission statements with the capacity of AI to read poetry, in this case Elizabeth Bishop’s ‘The Armadillo’, which GPT3 wrongly suggests centres on a car accident. I suggest that it is poetic interpretation, rather than management speak, that is real and testable – and hence less liable to replication. 2) I embed my account of the materials that GPT3 produces within a longer history of thinking about the nature of creativity in language, focusing on two particular moments: Wordsworth’s account of his emergence as a poet, and postwar French theory’s radicalization of skepticism about intention (‘there is nothing outside the text’). 3) I conclude with more concrete suggestions about how to respond to the challenges that GPT3 presents to university English. These include focusing more on ‘authentic’ assessments (which I describe in greater detail) and struggling to ensure that efficiency gains through AI are returned to students as opposed to the central university.






Essay available here:

Research statement

Background The essay asks how artificial intelligence will affect the teaching of literary studies in universities, and how literary studies should understand Large Language Models. I investigate how LLMs work, and how they can be understood against the background of theorizing in the discipline of literary studies over the last century. I focus here in particular on the work of Michel Foucault and Jacques Derrida. I also draw on approaches to creativity in literary thought, such as writing by William Wordsworth and Roland Barthes. The purpose is to suggest new ways forward for assessing and undertaking literary study in universities. Contribution The rapid emergence of Large Language Models presents significant challenges for literary studies in the university, from issues such as assessment to wide-scale philosophical questions about the nature of intention and literary value. My essay (1) situates these questions in a longer history of literary thought and practice, (2) examines the strengths and weaknesses of LLMs in relation to different kinds of discourse, and (3) identifies distinctive purposes for literary studies in an environment permeated by AI. The essay shows how effective LLMs are at producing management discourse and suggests that literary thought sustains human connection. Significance The 4000-word extended essay offers a significant contribution to literary studies, a field which is currently responding in a variety of ways to LLMs. My work speaks to essays on this topic recently published in e.g. the New Yorker, LARB, and elsewhere. My essay is published in the most prominent Australian literary magazine, the Sydney Review of Books, and is the first major public response to this issue of public and professional significance. The focus on how to teach in the discipline particularly advances existing conversations, as I seek to set the terms in a major public forum for Australian and international responses to LLMs in literary studies.


Sydney Review of Books

Place of publication



Sydney Review of Books

Usage metrics