If humans and AI disagree : A political approach to existential risk

This chapter introduces a distinctively political approach to the existential risks posed by artificial intelligence (AI) to all humans.Looking at how both humans and AI affect each other while neither side has sovereign control, the chapter explains the existential risks as a consequence of persisting disagreement between humans and AI.Depending on how relationships between humans and AI are constructed in a physical and social space, their disagreements may escalate into destruction and violent conflict but also stimulate development and learning.The suggested approach thus opens new opportunities to curb destructive conflicts between humans and AI before those conflicts get real.Generalising existing knowledge on the consequences of disagreements among humans, the chapter asks: How would an AI act in a political disagreement with humans if constructed in analogy with a state, a class, or a gender?Elaborating on different answers and testing procedures for this question, the chapter finds that human/AI relations should, to minimise existential risks for humans, mimic relations between genders.Humans should live close to AI and avoid the territorial segregation of states as well as the functional segregation of social classes in their relations with AI.