In a new blog post, Yoshua Bengio lays out his rationale for why we should be paying a lot more attention to the existential risks presented by future artificial intelligence