I agree, except I think stage 1 implies something more like “Group X acts badly in about 80% of examples of Situation Y.”
I think the criticism tends to be something like “white people” or “rich white men”, which I agree is very vague. I’m really keen we get better at predicting how likely EA is to screw up in particular ways by finding a better reference class.
I agree, except I think stage 1 implies something more like “Group X acts badly in about 80% of examples of Situation Y.”
I think the criticism tends to be something like “white people” or “rich white men”, which I agree is very vague. I’m really keen we get better at predicting how likely EA is to screw up in particular ways by finding a better reference class.