Is the Author Dead When the Machine Writes? On Literature in the Time of AI

Your current research project focuses on an aesthetic critique of generative AI. Could you tell us how it began?

The project originated in a master’s seminar on literary representations of AI I taught in 2024. In the classroom, we studied Death of an Author, a 2023 novella by the Canadian writer Stephen Marche, of which 95 percent of the sentences were generated by large language models. The title is an obvious play on Roland Barthes and the text a meta-detective story. In the seminar, some students saw the text as the death of art and most found it lacking in literary style. Yet there were some fascinating moments in our dialogue. I realised that it is not only authorship that is changing but also readership: what role does the human reader play in creating the meaning of these artificial texts? That shift in the communicative situation of literature became the starting point for my FRIAS project.

In what ways can a literary scholar confront the hazards of AI‑generated texts?

AI interfaces such as ChatGPT or Claude produce texts that often read perfectly well on the surface – grammatically sound, coherent across a certain length, often stylistically apt. How do we deal with these texts as critical readers? One could, of course, fight fire with fire and use AI to analyse AI-generated texts, and computational methods from the Digital Humanities are crucial for this. My FRIAS project adds to this a perspective rooted in aesthetics, poetics, and close reading. I am interested in the first-person experience of reading these texts and what aesthetic terms such as the sublime can tell us about that experience. Generally, we need cultural studies to counter hyped claims of Artificial Intelligence. This begins with the terminology itself. The term “generative AI” implies genuine creation, yet, as Kate Crawford notes in her Atlas of AI, artificial intelligence is arguably neither artificial nor intelligent. The discourse is saturated with metaphors that shape how we think about these systems. Critiquing these metaphors is a central concern of the emerging field of Critical AI Studies and an important task for cultural studies overall.

Can a machine‑produced text achieve the intentional communicative act of a human author?

For literary studies, a benchmark would be that a large language model produces a novel. Despite some claims, they cannot do so now and rather produce “slop” at that length. While a few authors of genre novels such as romance fiction say they use AI already, authors of literary fiction are even less likely to embrace it. Narratologist Jim Phelan has offered a useful distinction here: Large language models can recombine linguistic structures, but they lack rhetorical agency, the deliberate intention to create something new and influence a situation. I haven’t yet observed a large language model setting out on its own to write an epic poem or a Great American Novel.

Which literary genre do you consider most fertile for investigating AI in contemporary narratives, and why?

My own interest lies in crime fiction because it offers something that dominant AI narratives such as those of artificial general intelligence or superintelligence lack in their emphasis on utopian or dystopian near futures: a perspective anchored in the present. Crime fiction asks what is already having an impact on the now, rather than what might happen in 2027 or 2050. I am becoming more and more interested in how crime fiction across media depicts AI as detective or criminal and sometimes even victim, and how the specific conventions of the genre produce a different, grounded understanding of these technologies.

How has the interdisciplinary environment at FRIAS shaped your work?

FRIAS has been transformative for the project. For example, as I have realised during my time at FRIAS, genre serves as an ideal lens for AI-generated texts. Genre sits at the intersection of production and reception: it captures audience expectations while also describing what authors do when they write within or against conventions. What is more, AI cannot be confined to a single discipline. This semester almost half the fellows’ projects intersected with it in some way. Cross-disciplinary conversations, for example about legal, neo-colonial, and environmental aspects of AI, have sharpened my project’s focus. AI is here to stay, and while literary and cultural studies contribute a unique perspective through the analysis of narratives, metaphors, and genres, we absolutely need a transdisciplinary dialogue to further understand and critique its impact.