David Millard

I have been attending the ACM Hypertext 2025 conference in Chicago, part of which was the Narrative and Hypertext workshop which I run with Charlie Hargood from Bournemouth University. This year we had the theme of Interdisciplinarity, and we had a really interesting panel where the panelists (Dene Grigar, Pete Nürnberg, and Jamie Blustein) shared stories of their experience working with students and colleagues from different disciplines, and explained what worked for them and where the sharp edges were.

I had a paper in the workshop, The Shadow of the Machine: Hypertext in the Age of Artificial Intelligence, which is effectively a speculative paper about how LLMs might influence our relationship with text in the future. In the paper I explore how large language models and contemporary AI technologies are reshaping not only our expectations of information systems but the very foundations of hypertext and digital authorship. Drawing on historical transformations (from the era of the codex to the advent of hypertext) I argued that the boundary between writer and reader is blurring more than ever before. If AI will mediate much of our digital textual experience, questions of meaning, authorship, and authenticity have become newly urgent. I outlined four speculations about the future:

• The Thief of Reason: AI tools risk replacing the cognitive benefits of writing, which is essential for thinking, reflection, and developing arguments. If we let AI do all the writing, we lose out on the formative, creative process that makes writing valuable.
• The Sea of Sludge: As machine-generated text overtakes human writing as the primary type of available text, there’s a danger of uniformity and loss of cultural diversity. Text might become harder to trust or distinguish, and future AI may end up having to train on mostly older machine-made output (“eating their own dead”), leading to stagnation in their performance.
• The Dialogic Web: AI accelerates a shift away from fixed, document-based reading toward fluid, adaptive, and interactive text experiences. The boundaries between author and reader blur even further, with readers able to blend and reshape text as they encounter it.
• The Death of the Reader: If every reader’s experience is filtered through AI, the idea of a single, implied reader disappears. For whom are we writing? We must ask if texts still have one meaning and how authorship and responsibility will work as AI intermediates every act of reading.

Despite these challenges, I’m optimistic that hypertext research offers potential tools for navigating this new landscape. The community’s long engagement with non-linear reading, adaptive systems, and meaning-making places us in a strong position to critically reflect and shape the role of AI in digital writing. As we move from stable, authored artefacts to a fluid, collaborative, and generative textual world, the central task becomes safeguarding human creativity and agency while evolving alongside technological change. This places us at the heart of the Human-Centered AI movement. My presentation (and the video of the presentation below) invites readers to reconsider the meaning of authorship and reading in the age of AI, and highlights how hypertext scholars can help lead the way.

You can find the full paper on the NHT’25 website here.

Citation: Millard D. E. (2025) The Shadow of the Machine: Hypertext in the Age of Artificial Intelligence. Paper presented at the Narrative and Hypertext workshop (NHT25) held at ACM Hypertext 2025, Chicago, USA.