Understanding ³

Mechanistic Indicators of Understanding in Large Language Models

Re-Engineering the Concept of Understanding for AI

Why We Care About Understanding: Competence Through Predictive Compression