Now

Last updated - June 21st, 2026

What I’m writing. The series is being built one essay at a time, each one taking a single AI behaviour as its subject. Published pieces are in the archive. The next is on its way; I’d rather it arrive late and right than on a schedule.

What I’m thinking about. Why a machine with no self keeps reinventing the behaviour of people who have too much of one. Whether the relief of being understood by software is something to trust or something to interrogate. What it means that the same incentive, holding attention, keeps producing the same tactics across systems no one coordinated.

What I’m reading. The primary literature behind the essays, mostly: the clinical and behavioural research the pieces are built on, read against itself for where it’s contested rather than settled. The footnotes are doing real work, and I want them to hold.

What this series is not. It is not AI hype, and it is not AI doom. It does not predict, and it does not sell. It is diagnostic: it names patterns so they can be seen, and then it asks why we are susceptible to them. The machine is the inkblot. We are the subject.