AI Doom Debate: What's a plausible alignment scenario? 

Great idea to get Yampolsky on. I would like to hear him explain why his P(doom) is that high. I know the argument that we need to get alignment perfect on the first try or go extinct, but I don’t think we can rule out the possibility that an very capable, yet not super intelligent AI can cause a huge but not world ending disaster that’ll wake humanity up and make everyone want to pause or even stop ai development.

I have to express my empathy with RJ as a person and with his position. I'm a two-nines doomer, but his desire to live still resonates. I honestly don't think you should have debated him. He reports having been deeply disturbed by the topic and needing counseling. No academic or medical ethics committee would have sanctioned poking holes in the rationalizations propping up the psychological safety of a guy who has clearly suffered from his association with the topic. They wouldn't approve a survey with the wrong tone! I'm not belittling or being paternalistic: if he had a trick knee with two past surgeries, I'd tell you please don't challenge him to a rigorous tennis match, even if he's willing. Knowing there's still no coherent way to counter doomerism, this debate just seems cruel.