Thanks to Laura Dietz for having me at UCL’s Department of Information Studies last week to talk about The Pleasures of Reading in the Age of AI.

The recording is now up on their YouTube channel if you want to tune in.

The talk was an excuse to think out loud with some smart and interesting folks about question I’ve been sitting with for a while: what happens to reading — particularly reading for pleasure — when AI starts showing up at every stage of the encounter with a text?

Reading is one of the most time- and attention-intensive things we do voluntarily, and AI is quickly altering our relationship with reading and speed. Gloria Mark’s research shows that the average time a knowledge worker spends on a single task before switching has dropped from around 150 seconds two decades ago to roughly 47 seconds today.

What AI seems to be doing is accelerating a shift that Louise Rosenblatt’s theory on transactional reading captures well — from aesthetic reading (where the reading experience itself is the point) toward efferent reading (where you’re extracting information and moving on). That’s not inherently bad, but almost all of AI’s reading affordances — “summarise this”, “explain this”, “give me the key points” — are oriented toward extraction. And I wonder what we quietly lose when we optimise away the friction.

I also raised the topic of something I’ve noticed working with students: what seems to be a reduced tolerance for ambiguity. The “just tell me the answer” impulse has always existed, but AI’s tendency to respond to ambiguous, or contested questions with the same confidence it uses for things with clear answers is, I think, a subtle but real epistemological problem. It’s something we’re looking at directly in our current Think Lab project with Wikipedia — specifically how AI platforms attribute and mediate Wikipedia content, and what it means that people are increasingly treating that as a substitute for visiting other more direct sources.

One thing that surprised me most in our live poll was how many people are using AI during reading — for clarification, discussion, or as a kind of book club proxy when you wanted that experience of sharing a reading experience with someone (or something).

I’m cautiously optimistic about AI as a thinking partner in reading — tools like Kindle’s X-Ray or “Ask the Book” feel like fairly useful extensions of the reading experience rather than replacements for it.

I wonder if this AI-driven acceleration of the reading experience might in turn spark a revival of the slow reading conversation from circa 2010.

If any of this is your territory — reading, attention, AI, digital habits — I’d love to hear from you.

You can check out the full video via the UCL Department of Information Studies YouTube Channel: