I encountered this over the Summit weekend. 1.5 hour lunch with a couple of people who could not stop solving the Friendly AI problem.
Out of curiosity, what is the most promising suggestion regarding Friendly AI that you’ve heard from someone who had probably spent fewer than 48 hours thinking about it?
Out of curiosity, what is the most promising suggestion regarding Friendly AI that you’ve heard from someone who had probably spent fewer than 48 hours thinking about it?