Coherent extrapolated volition (alignment target) - History — AI Alignment Forum