Prof Russell talks about what goes into training ChatGPT-like AI bots and how because of the self-learning algorithms, can have a catastrophic effect on humanity. He elaborates on how we need to understand the importance of getting these systems to behave themselves so that there is no real significant risk to the public.