Statistically responsible artificial intelligences

Abstract As artificial intelligence (AI) becomes ubiquitous, it will be increasingly involved in novel, morally significant situations. Thus, understanding what means for a machine to responsible is important ethics. Any method ascribing moral responsibility AI must intelligible and intuitive the humans who interact with it. We argue that appropriate approach determine how AIs might fare on standard account of human responsibility: Strawsonian account. make no claim our either only one worthy consideration or obviously correct approach, but we think preferable trying marry fundamentally different ideas (i.e. AI, humans) into single cohesive Under framework, people are when they appropriately subject particular set attitudes—reactive attitudes—and under conditions machines this same attitudes. Although traditionally applies individual humans, plausible entities not possess these attitudes candidates framework. conclude weak never responsible, while strong right emotional capacities may responsible.