#64 – Michael Aird on Strategies for Reducing AI Existential Risk
Hear This Idea
00:00
The Importance of Defense in Depth
So first you had the idea of an international multilateral monitoring and enforcement agency a bit like what the IAEA is to nuclear security before AI. The second thing was the thought that the US or US led coalition might use geopolitical influence to just stop unsafe AI development. And then the final thing was a modeling through story where we just have lots of overlapping uncorrelated safety measures which collectively turn out to be enough to avoid catastrophe. All of these would also have to involve craploads of alignment research.
Transcript
Play full episode
