Don’t look at me to resolve that conflict. I think moral extrapolation is unlikely to output anything coherent if the reference class is sufficiently large to avoid the objections I raised above. And I can’t think of any other plausible candidate to produce Friendly instructions for an AI.
Don’t look at me to resolve that conflict. I think moral extrapolation is unlikely to output anything coherent if the reference class is sufficiently large to avoid the objections I raised above. And I can’t think of any other plausible candidate to produce Friendly instructions for an AI.