Domesticating Artificial Intelligence

Abstract For their deployment in human societies to be safe, AI agents need aligned with value-laden cooperative life. One way of solving this “problem value alignment” is build moral machines. I argue that the goal building machines aims at wrong kind ideal, and instead, we an approach alignment takes seriously categorically different cognitive capabilities between agents, a condition call deep agential diversity . Domestication answer similarly structured problem: namely, how integrate nonhuman animals lack agency safely into society align behavior values. Just like animals, genuinely agency; just might find ways train them nevertheless assist us, live work among us – “domesticate” them, other words. claim domestication does well explaining many our intuitions worries about deploying social practices.