We can’t ensure the safety and stability of AI as long as lives and fortunes are tied up in tolerating or diluting AI’s repercussions.