I once counted several dozens of the ways how AI can cause human extinction, may be some ideas may help (map, text).
See also ‘The Main Sources of AI Risk?’ by Wei Dai and Daniel Kokotajlo, which puts forward 35 routes to catastrophe (most of which are disjunctive). (Note that many of the routes involve something other than intent alignment going wrong.)
I once counted several dozens of the ways how AI can cause human extinction, may be some ideas may help (map, text).
See also ‘The Main Sources of AI Risk?’ by Wei Dai and Daniel Kokotajlo, which puts forward 35 routes to catastrophe (most of which are disjunctive). (Note that many of the routes involve something other than intent alignment going wrong.)