WHAT IS IT TO SOLVE THE ALIGNMENT PROBLEM? (2 HOUR READ)

Summary from the TLDR newsletter

Solving the alignment problem means avoiding a bad form of AI takeover while building the dangerous kinds of superintelligent AI agents and gaining access to the main benefits of superintelligence from those agents.