At CAAIE I lead the programme and teach the curriculum for AI in health and administration — two sectors where the stakes are highest, the data is most sensitive, and adoption is slowest. That combination is exactly why I teach there.
Who sits in the room
Not engineers. Nurses, administrators, clinic managers, public-sector specialists — people who run the systems society actually depends on. They do not need to build models. They need to judge them: what a model can do, where it fails, what may never leave their infrastructure, and which vendor promises are physics and which are marketing.What the curriculum does
We start from their processes, not from the technology. Documentation, triage, correspondence, reporting — we take real workflows and rebuild them with AI in the loop, always with a human owning the outcome. By the end, participants have automated something real from their own daily work. That changes more minds than any lecture.Why literacy beats tools
Tools change quarterly; judgment does not. An administrator who understands data protection, model limits and prompt design will make good decisions with whatever tool arrives next year. That is the actual product of education — durable judgment, not software training.Teaching is also how I stay honest: nothing exposes a shallow understanding faster than a room full of practitioners asking 'but how, exactly?' I teach to give that answer — and to keep having to.