Writing
Essays about shipping AI to people who can't absorb a wrong answer, and about the judgement calls that survive after the demo works.
Newest first, from Medium.
what readers said
Such a sharp, clear-eyed analysis of the challenges we're all grappling with.
This is a sensational write-up
Perfect if you're currently navigating the messy, magical reality of AI implementation
One of the few AI articles that I felt was worth reading in recent times.
on AI, product and trust
The agent loop is the easy part; knowing when to stop it is the work. The thinking underneath this site's /loops page.
Models have infinite knowledge and no habits. On writing the fixes down so a correction survives the session.
Capability is not adoption. What has to be true before someone lets a model act on their behalf.
Deploying to first-generation university applicants, people lost in visa paperwork, job seekers in a brutal market. What changes when a hallucination closes a door permanently, read against the EU AI Code of Practice and a run of real failures.
A city did everything the responsible-AI playbook asks for and the system still failed. Where governance-by-checklist breaks.
Field notes from a week with the people actually deploying this, and the gap between the conference talk and the rollout.
podcast
Building a Career Co-pilot for Disadvantaged Students: How Zero Gravity Bridges Knowing and Doing
I went on Just Now Possible, hosted by Teresa Torres, to talk about building AI that closes the gap between knowing what to do and actually doing it, for students without the network that usually supplies the answer.
earlier
- Learning Anxiety in the Age of Artificial Everything · 2025-05-23
- The Walled Garden: How Zuckerberg Tried To Build His Own Internet · 2025-04-22
- What Finding Nemo Taught Me About Tech for Good · 2025-04-11
- The Algorithmic Gentrification of Your Sunday Roast · 2025-04-04
elsewhere
I studied AI governance and alignment with BlueDot Impact.