Human-friendly goals

1. Direct AI Alignment vs Indirect AI Alignment (Prompt Engineering Secrets)