Eliciting Latent Knowledge

Deep Double Descent

Chinchilla’s Wild Implications

Intro to Brain-Like-AGI Safety

Gradient Hacking: Definitions and Examples

An Investigation of Model-Free Planning

Discovering Latent Knowledge in Language Models Without Supervision

Toy Models of Superposition

Imitative Generalisation (AKA ‘Learning the Prior’)

ABS: Scanning Neural Networks for Back-Doors by Artificial Brain Stimulation

Least-To-Most Prompting Enables Complex Reasoning in Large Language Models

Two-Turn Debate Doesn’t Help Humans Answer Hard Reading Comprehension Questions

Low-Stakes Alignment

Empirical Findings Generalize Surprisingly Far

Worst-Case Thinking in AI Alignment

How to Get Feedback

Public by Default: How We Manage Information Visibility at Get on Board

How to Succeed as an Early-Stage Researcher: The “Lean Startup” Approach

Become a Person who Actually Does Things

Working in AI Alignment