Smart Contracts as Credible Commitments for Trading with Scheming AIs
Epistemic status: medium confidence. I believe that smart contracts are promising mechanisms for trading with…
Category: Opinion
A “Lay” Introduction to “On the Complexity of Neural Computation in Superposition”
This is a writeup based on a lightning talk I gave at an InkHaven hosted…
Reflectively stable consequentialists are expected utility maximisers
Epistemic status: Theorem.When we look at the world we see a repeating pattern: people, corporations,…
Savage’s Axioms Make Dominated Acts EU Maxima
A common coherence defense of EU is that it blocks money pumps and exploitation. Yet…
Does your AI perform badly because you — you, specifically — are a bad person
Claude really got me lately.I’d given it an elaborate prompt in an attempt to summon…
Cost vs. Profit Center Mindset
In 2016, Volkswagen became the world's largest automaker, overtaking Toyota by number of cars sold.…
Pando: A Controlled Benchmark for Interpretability Methods
TL;DR: Pando is a new interpretability benchmark with 720+ fine-tuned LLMs carrying known decision rules…
Opus 4.7 Part 2: Capabilities and Reactions
Claude Opus 4.7 raises a lot of key model welfare related concerns. I was planning…
I built a semantic search engine for LessWrong
For a while now, I've wanted something like Connected Papers for LessWrong & the broader…
The Right Answer to “Can You Keep a Secret?”
Please do not talk about this. I was asked to keep this private. And I'm…