The speaker discusses the concerns related to AI causing harm, leading to various potential issues such as unemployment, spreading fake news, and creating harmful pathogens. One proposed solution is to align AI's values with human morality, but the concept of alignment is complex due to diverse human moralities. The speaker expresses a deep interest in alignment research and its limitations, highlighting the challenge of aligning AI with differing human values.
It seems obvious that moral artificial intelligence would be better than the alternative. But psychologist Paul Bloom of the University of Toronto thinks moral AI is not just a meaningless goal but a bad one. Listen as Bloom and EconTalk's Russ Roberts have a wide-ranging conversation about the nature of AI, the nature of morality, and the value of ensuring that we mortals can keep doing stupid or terrible things.