I think there should be a serious discussion about AI risks, but it has to include the counterfactual: what are the risks of NOT building, both in terms of a world of asymmetric capabilities and what that means for our societies in terms of security and growth (i.e., the human