Arianne Burrow
What does it mean to create meaning in literary analysis in the age of Artificial Intelligence?
Since the emergence of generative AI tools such as ChatGPT, the process of interpreting texts is no longer exclusively human. In the Digital Humanities, this raises a significant question: if interpretation is increasingly mediated by AI systems, can meaning still be said to emerge from the interaction between reader and text, or is it shaped by the structures that guide AI outputs?
This project explores that question through the lens of Roland Barthes’ influential essay ‘The Death of the Author’ (Barthes). Barthes famously argues that ‘the birth of the reader must be at the cost of the death of the Author,’ positioning meaning as something created through the reader’s engagement with the text rather than determined by authorial intention (Barthes 130). Meaning, therefore, is not fixed but emerges through interpretation. However, this raises a contemporary problem: What happens when the ‘reader’ is no longer human, but AI?
A useful starting point is Kieran O’Halloran’s adaptation of ‘digital assemblages’ to an AI/digital context, developed in ‘Digital Assemblages with AI for Creative Interpretation of Short Stories’ (657). O’Halloran proposes a pedagogical model in which literary interpretation is produced through an assemblage, a network of elements including critical theory, corpus analysis tools, and AI systems such as ChatGPT. In this model, AI contributes to interpretation by extending human analysis in creative and unexpected ways.
Crucially, O’Halloran argues that ChatGPT can be used ‘aleatorily’ to extend interpretation, that is, as a source of unpredictable or random insights (663). He suggests that ChatGPT can recover overlooked themes, synthesise large datasets, generate ‘hallucinations’ to inspire creative insights, and reveal hidden textual patterns. In this sense, AI appears to function as a productive collaborator, expanding the interpretive possibilities available to the reader. This project builds on O’Halloran’s approach by applying ChatGPT to Sylvia Plath’s poem ‘You’re’, using differently framed prompts to test how ChatGPT-generated interpretation shifts in response to prompt bias (Plath).
While this project does not reject O’Halloran’s claim that AI can enhance interpretation, it challenges the assumption that ChatGPT can operate in a genuinely aleatory or unpredictable way. Instead, it argues that ChatGPT outputs are fundamentally shaped by prompt bias: the framing, structure, and assumptions embedded in user input. Rather than producing meaning independently, AI systems respond to and reinforce the interpretive directions encoded in prompts.
This shifts the role of AI from a creative co-interpreter to a mediating force that steers interpretation itself. If prompts guide what AI ‘sees’ in a text, then interpretation is not open-ended but directed. In this sense, meaning may not be freely produced through reader-text interaction, as Barthes suggests, but instead steered through prompt design.
The importance of this extends beyond literary studies. Research by Shu et al. demonstrates that prompt bias in AI-generated texts can have measurable real-world effects (Shu et al.). While the study also compares AI-generated summaries with sources such as Wikipedia, its findings on prompt framing are particularly significant. In a survey of 1,912 participants, individuals were exposed to AI-generated summaries of historical events framed in neutral, liberal, or conservative ways. The results showed that liberal-framed outputs increased liberal attitudes across all groups, while conservative framing reinforced conservative views, primarily among those already aligned with them. This study highlights that AI-generated texts do not merely reflect information but actively shape interpretation and opinion. The implications for literary analysis are significant: if prompts can influence political beliefs, they can also influence how texts are read and understood. Interpretation, therefore, becomes not just a matter of reading but of prompting.
This project positions itself within this tension. While Digital Humanities scholarship often celebrates AI as a tool for expanding interpretation, it is equally important to recognise how its outputs are constrained. If AI mediates the relationship between reader and text, then the interaction Barthes describes may be fundamentally altered, or even rendered impossible. In this sense, this project contributes to ongoing debates about authorship, interpretation, and AI literacy by questioning whether meaning in AI-assisted analysis is genuinely created or structurally guided.
Initial user feedback on this project emphasised that the presentation was overly dense and moved too quickly between complex ideas, limiting audience engagement. In response, I streamlined the structure of the project, reducing the amount of information presented at once and focusing more clearly on the central argument about prompt bias.
Additionally, I refined the pacing and visual elements of the presentation to foreground the literary text itself, allowing space for appreciation of the primary material rather than overwhelming it with analysis. This shift ensured that the project remained grounded in literary studies while still engaging critically with digital tools.
Ultimately, this project asks whether AI challenges or reinforces Barthes’ vision of interpretation. If prompts act as a middleman –skewing the author’s text and steering the reader –then the interaction between them is no longer direct. In that case, meaning may not emerge through reading alone but through the structures that guide AI interpretation.
And so, when AI mediates literary analysis, is meaning ever truly made?
Bibliography
Barthes, Roland. ‘The Death of the Author’. Authorship: From Plato to Postmodern: A Reader, edited by Sean Burke. Edinburgh University Press, 1995, pp. 125-30, doi-org.eux.idm.oclc.org/10.1515/9781474465519-017.
O’Halloran, K. ‘Digital Assemblages with AI for Creative Interpretation of Short Stories’, Digital Scholarship in the Humanities, vol.39, no.2, 2024, pp. 657-689, doi.org/10.1093/llc/fqad050 .
Plath, Sylvia. ‘You’re’. Ariel. Faber and Faber, 1965, p. 51, poetryfoundation.org/poems/49010/youre.
Shu, Matthew, et al. ‘How Latent and Prompting Biases in AI-Generated Historical Narratives Influence Opinions’, PNAS Nexus, vol. 5, no.3, 2026, pgag022, doi.org/10.1093/pnasnexus/pgag022 .
Cite this page:
Burrow. Arianne. 'The Death of the Author and The Birth of the AI Reader'. Cream of the Slop. version 1.0, Digital Humanities for Literary Studies 2025-26, University of Edinburgh, 10 Apr. 2026, https://blogs.ed.ac.uk/dh2025-26/.
