ABSTRACT
One of the primary benefits to conducting any type of comparative research is the insight gained from the origination field through the process of exploring the destination field. Though significant hand-wringing has gone into opining on how to avoid an artificial intelligence (AI) planetary takeover, fewer moments have been spent on knowing what to teach them. It is easy to say that we should teach AI to be ethical and philanthropic, but we don’t have an excellent record in teaching other humans to be ethical and philanthropic. If we have to send repeated email solicitations to long-time donors for contributions to our annual operating fund, where do we start training beneficent AI?
This chapter provides a framework for distilling such learning in parallel with inquiries about human intelligence. Questions regarding conditions for moral agency, the power of explicit and implicit rules, the influence of peers, and the emergence of unanticipated consequences are asked of both human and artificial learning. Further, philanthropy provides an ideal context for asking such questions since the host of possible human motivations are still hotly debated in the social sciences. The resulting insights provide not only practical advice on how to train AI but also move the conversation forward on broader epistemological and ontological questions of intelligence, regardless of how the intelligence came into being.
