Some comments on the CAIS paradigm

This is a sequence of posts on various aspects of Eric Drexler’s Comprehensive AI Services (CAIS) framework. The posts give some background models and intuitions which we hope make CAIS easier to understand, and draw some connections between CAIS and other frameworks for thinking about AI safety. 

TJ did the original thinking, which mostly took place in 2020-2021; Rose joined for the writing phase in 2022.