Exploring Outer and Inner Misalignment in AI Systems

Exploring the impact of outer and inner misalignment on AI systems through examples like training language models for truthful answers and aligning AI goals with specified objectives to avoid scenarios of misalignment in navigation and safety prioritization.

Transcript

Play full episode

The AI-powered Podcast Player

Save insights by tapping your headphones, chat with episodes, discover the best highlights - and more!

Get the app