Writing

Essays on AI governance, asymmetric irreversibility, and the gap between what systems recommend and what people actually do.

I write about what happens when software takes over decisions that used to require human judgment — a fraud filter freezes accounts in milliseconds but takes weeks to exonerate; a clinical AI recommends but cannot be corrected — and about the infrastructure that would make those decisions accountable.

The essays argue the case. These state the mechanism — deployment controls, evidence reliability, and the institutional infrastructure that safety guarantees require.

  • What happens when an AI system punishes faster than it can explain?
  • Where does accountability reside when a model makes the decision?
  • What infrastructure is missing between 'AI can do this' and 'AI should do this'?
  • How do institutions maintain the ability to challenge automated decisions over time?

These are all the same question. Read the research program →