There’s a particular flavour of déjà vu that comes with walking into Christ Church College, Oxford to speak at a Mexican symposium. The dome of the Radcliffe Camera in one eye, the Ángel de la Independencia in the other, and a room full of people who, like me, ended up here by way of a path that didn’t appear on any prospectus. The visit reminded me of my own time in Oxford and the memories floded in. Anyway, I was very plesed to have been invited to be a keynote speaker at the XXIII Symposium of Mexican Students and Studies in the United Kingdom and I came away convinced that the most useful part of the day hadn’t actually happened yet. More on that below.
What I talked about
The talk was called A Model Life — partly a pun, partly a confession. The arc was simple: how does a physicist end up running AI and innovation work across industries that, on paper, have nothing to do with each other? The short answer is that physics never really lets you go. It just changes costume.
I traced the throughline from modelling physical systems, where you have the luxury of laws, symmetries, and conservation principles, to modelling human and organisational systems, where you have none of those things and still have to ship something by Friday (or Monday, I don’t like releasing before a weekend, ask my teams!). The interesting move isn’t the maths. It’s learning when to stop trusting the model. Physicists are trained to know the regime of validity of our equations; most of the failures I’ve seen in applied AI come from people who never learned that habit.
From there, the talk pivoted into where this is all heading: agentic systems, multi-agent orchestration, and the awkward truth that “AI strategy” in most organisations is still a wishlist dressed up as a roadmap. I argued that if an AI deployment doesn’t change behaviour, improve a decision, reduce cost, or unlock something previously impossible, it’s noise. The room seemed to agree. Or they were being polite. Hard to tell at Oxford.
The career thread underneath all of it: there’s no canonical path from a physics PhD to leading technology work, and that’s the point. Lateral moves compound. Optionality is a real asset class. And the discipline of thinking from first principles, which physics drills into you whether you like it or not, turns out to be the single most portable skill in a market that retools itself every eighteen months.
What I didn’t get to talk about
Q&A is always too short. I left the stage with the feeling that the conversations I actually wanted to have were the ones queued up in people’s heads that never made it to the microphone. The student working on something genuinely interesting who didn’t raise a hand. The researcher with a half-formed idea they weren’t sure was “ready”. The person who had a question but framed it as a comment.
So I’m doing something about it.
I am proposing the OxMex Society — who organised the symposium and deserve credit for an exceptionally well-run two days — to open a call for submissions to a very simple competition. It’s deliberately low-friction and format-agnostic. I hope to be able to provide more information about that once we have something for solid.
Thanks to the Oxford Mexican Society, MexSocUK, the Embassy of Mexico in the UK, and Christ Church for hosting. Two days well spent.
