@kevinnbass Science used to be done by a small number of superintelligent obsessives, but now is done by a large number of moderately intelligent careerists.
@bindureddy One piece of circumstantial evidence that we might be in trouble is the glibness and unseriousness of AI boosters, and their terrifyingly naive utopianism. The ones who are worried seem like the grownups (Hinton, Bengio).
@S_Conradi Very nice, going to see how you do this in numpy. I found that spawning particles according to some weighting function (e.g. a 2D image) can give nice results: https://t.co/a4APyWw9mm
@pwang How would consensus over training materials be achieved? I wouldn't accept the pronouncements of an AI that my political adversaries had trained.