David Robinson / The Atlantic: David Robinson, ex-OpenAI safety and policy: SV lacks a safety-centric culture; labs must study other fields' safety approaches; time for trial and error's over — The industry's approach to safety will guarantee more failures unless something changes.