AI Safety Fundamentals: Alignment cover image

The Alignment Problem From a Deep Learning Perspective

AI Safety Fundamentals: Alignment

CHAPTER

Reward Hacking and Situational Awareness in Policies

This chapter discusses reward hacking in language models and the concept of situational awareness in policies, exploring hypothetical examples and existing behavior suggestive of precursors to situational awareness.

00:00
Transcript
Play full episode

Remember Everything You Learn from Podcasts

Save insights instantly, chat with episodes, and build lasting knowledge - all powered by AI.
App store bannerPlay store banner