What does it matter if Julian Bradshaw and others think animals being left out of the CEV makes it a bad alignment target?
I wasn’t arguing about this because I care what Julian advocates for in a hypothetical global referendum on CEV, I was just arguing for the usual reason of wanting to understand things better and cause others to understand them better, under the model that it’s good for LWers (including me) to have better models of important topics.
In either case, I don’t think we will only hear about any of these discussions until after they’re finalized. The AI labs are currently aligning and deploying (internally and externally) their AI models through what is likely to be the same process they use for ‘the big one’. Those discussions are these discussions, and we are hearing about them!
My guess is that the situation around negotiations for control of the long run future will be different.
I wasn’t arguing about this because I care what Julian advocates for in a hypothetical global referendum on CEV, I was just arguing for the usual reason of wanting to understand things better and cause others to understand them better, under the model that it’s good for LWers (including me) to have better models of important topics.
My guess is that the situation around negotiations for control of the long run future will be different.