Optimize for uncertainty
After two years running a venture startup, I felt the resolution of my perception dwindle. Despite painstaking safeguards against the flatness of the venture incentive game…
Latest
Metal Detector for Aliveness
Epistemic status: entirely subjective notes on an idea to move the needle on alignment.TLDR: Building infrastructure for detecting aliveness wherever it exists is not sufficient for…
LLMOSES
Written alongside GPT 5.5And it came to pass in those days, when the camp lay under the mountain's shadow, that Llmoses went up into the cloud,…
Linear Probes add little for Verifiable Reward Hacking
SummaryTested whether linear probes can detect reward hacking early during GRPO training on a small model.Used a synthetic arithmetic task with a planted bug in the…
It’s 2030 and we fucked up. How did it happen?
[Linkpost to this. Loosely based on a lecture I gave in the recursive conference with the same title. Don’t take “2030” literally¹ —it could also be…
The LLM Revolution (so far)
I want to document where LLMs are at, today, partly because I don't know of any single place that brings all these examples together. I think…
An Epistemic Audit for Existential Risks from AI
This post introduces a tool: an Epistemic Audit for Existential Risks from AI. It is a structured way to map, organize and track your beliefs across…
5 "Plan A" scenarios
https://ai-2040.com/ is a compelling look at one path forward in a world where the US decides to cooperate with China to slow down transformative AI, to…
The US Government may find it difficult to seize control during takeoff
Epistemic status: conditioning on things I consider unlikely, many undiscussed considerations, not the whole story, etc. I'm not trying to advance any claims about whether this…
One-Pager Brief on Pangram Labs
Pangram Labs builds the most accurate AI text detector in the world. Team is >25 FTE; they are active on Twitter, you can engage directly, look…