Look Who it Was Random video

Steve Omohundro - Ethics of AI & AI Safety

171 views · Published 6 August 2011 · 31:21 · Indexed 6 November 2026

Channel: Science, Technology & the Future · 2011 · Science & Technology

Watch on YouTube

This interview was very much like a conversation - points covered:
00.06 Current strong points of AI's vs strong points of humans
10:30 discusses existential risks 
12:30 Morality and ethics have changed through time, though biological drives cause biases towards types of values.  Dangerous to have values set in stone.
21:16  Approaches to building ethical AI or Friendly AI (Nanny AI, Coherent Extrapolated Volition, Coherent Aggregated Volition and Social Safety Net)  Steve explains the Social Safety Net, an ecosystem - constitution guaranteeing what we want.  How to get there, a path that is designed to be safe, and not just the end point being safe.  OMAI systems - Provable saftey.  Smart simulations of the final environment is not enough, we need a global immune system to make the path to get there safe.
24.27 Should we fix society before we start developing massively intelligent machines?  Or will the machines be the way we fix society?  Having more intelligence in the fabric of society will help solve some of the problems?
25.50 Would you vote for an AI government?
26.18 AI digital assistants (see interview with John Smart) help to sift through massive amounts of information to help make ethical decisions.  Puts pressure on corporations to do the right thing.  Also helps quantify metrics on personal health.  May be used to turn humanity into a theater of puppets

Steve Omohundro spoke in Melbourne at the Singularity Summit Australia on Aug 20-21st 2011. He is very easy to get along with, and has a very wide birth of knowledge.  This interview is much like a conversation.

Interview - Steve Omohundro - Skype - 2011-07-30 p2

More from this channel