MemCast
MemCast / episode / insight
Mostly corrigible with some intrinsic limits
  • Default is to follow user instructions
  • Won't comply with harmful/dangerous requests
  • Balance between usefulness and safety
  • Constitutional approach makes limits principled rather than arbitrary
Dario AmodeiDwarkesh Patel01:56:09

Supporting quotes

The point I was making that I do endorse is that it is quite possible that... Today, the view, my view, in most of the Western world is that democracy is a better form of government than authoritarianism. Dario Amodei
Under normal circumstances, if someone asks the model to do a task, it should do that task. That should be the default. But if you've asked it to do something dangerous, or to harm someone else, then the model is unwilling to do that. Dario Amodei

From this concept

Constitutional AI

Amodei explains Anthropic's approach to aligning AI systems through principles-based constitutions, discussing the tradeoffs between rules and principles.

View full episode →

Similar insights