I teach the way I write, which is by showing the mechanism and letting the room draw its own conclusion. Three things live here.
Reading the Fuzz
A sixty-minute guest session for cyberpsychology students at ATU on how to tell when an AI is going off the rails. It isn’t about my research or my tools; it’s a set of habits for spotting the moment a fluent answer stops being a true one. The centrepiece is a live trap. The room is sent to the same chatbot, fed a false premise about a real researcher’s work, and asked to see which of them notices that the model didn’t push back. When I built the session in August 2026 the trap still caught one of the major models every time, and it’s a far better teacher than any slide about hallucination. A presenter’s run book and a ten-slide deck exist for anyone who wants to run it.
Trust, cybertrust, and AI
An interactive exploration, not a lecture, built for a master’s cohort in Applied AI. Three acts: what trust is as a psychological state; what I call cybertrust, which is trust under mediated conditions, when the thing you’re trusting arrives through a screen with no body behind it; and then the question of what that means when the thing on the other side is a model. Forty-eight references, every one with a verified DOI, because a session about trust that cited a paper that didn’t exist would be its own punchline.
When Machines Might Matter
A podcast script I wrote for a cyberethics module, on AI rights, consciousness and the gap in the law, drawing on the Universal Declaration of Sentient Rights. I’m turning it into a series of short films, each part under a minute and labelled on screen, made with Everygen and voiced by a synthetic English voice rather than mine, which was the honest choice for a piece about whether a synthetic voice could ever be someone. They’ll be linked here as they’re cut.
If you want any of these for a room of your own, ask.