johnswentworth

The Plan − 2024 Update

The Field of AI Align­ment: A Post­mortem, and What To Do About It

[Question] What Have Been Your Most Valuable Ca­sual Con­ver­sa­tions At Con­fer­ences?

The Me­dian Re­searcher Problem

Three No­tions of “Power”

In­for­ma­tion vs Assurance

Min­i­mal Mo­ti­va­tion of Nat­u­ral Latents

Values Are Real Like Harry Potter

We Don’t Know Our Own Values, but Re­ward Bridges The Is-Ought Gap

Why Large Bureau­cratic Or­ga­ni­za­tions?

… Wait, our mod­els of se­man­tics should in­form fluid me­chan­ics?!?

In­ter­op­er­a­ble High Level Struc­tures: Early Thoughts on Adjectives

A Ro­bust Nat­u­ral La­tent Over A Mixed Distri­bu­tion Is Nat­u­ral Over The Distri­bu­tions Which Were Mixed

What is “True Love”?

Some Unortho­dox Ways To Achieve High GDP Growth

A Sim­ple Toy Co­her­ence Theorem

A Solomonoff In­duc­tor Walks Into a Bar: Schel­ling Points for Communication

(Ap­prox­i­mately) Deter­minis­tic Nat­u­ral Latents

Dialogue on What It Means For Some­thing to Have A Func­tion/​Purpose

3C’s: A Recipe For Mathing Concepts