· 1 min read

Subfields of AI safety

I’m trying to distinguish subfields of AI safety. Which parts are wrong?

  • Alignment (do what we want)
  • Control (don’t do what we don’t want)
  • Ethics (don’t make society worse)
  • Policy (policy)
  • X-risk (don’t destroy the world)
  • S-risk (don’t make suffering AI)

View original