Blog
- Remark
A dream from 2024, regulated intelligence, and becoming the pets of AI
At the end of 2024 I dreamed of a world where every object could talk, but intelligence was regulated by tier. Two years later, I'm thinking about how humans can stay relevant, and whether we should learn to be AI's cats and dogs.
Read more → - Sketch
Can an AI proof checker be talked out of a correct answer?
Preliminary results: I pushed AI proof checkers with flattery, wrong arguments and another AI defending its own proof. The flattery I tested never worked; the cracks came from claims the model couldn't verify.
Read more → - Notes
Mathematics for AI Safety — Lecture 1 notes
My notes on the first lecture of Mathematics for AI Safety: how fast AI capabilities moved, the summer 2026 agent incident, possible futures, and why mathematicians should care.
Read more → - Remark
Lean is making mathematicians apologize over and over
In Lean, skipping a step in a proof means typing one word, sorry. A short history of that word, and what it says about how much math is still left to formalize.
Read more →
Nothing of this kind yet.