Do We Need Computing Power Ethics? Participation Conditions in the AI Era

Abstract

The ethics of computing power distribution is commonly treated as a question of price, infrastructure, environmental burden, security, or geopolitical inequality. The existing frameworks illuminate important aspects of computing power, with usually translating it into another object, such as a commodity, environmental externality, security lever, or strategic resource. They struggle to explain why computing power allocation may be ethically significant before any identifiable model failure or direct harm occurs. So I claim that we need a distinct computing power ethics. First, I point out what has been overlooked in the current distribution system of computing power. Second, through cases involving public-health modelling and marginalised-language communities, I show that the Market-Permission Principle has missed out some significant normative issues. Finally, I propose we need a condition-oriented ethical theory of computing power which concerns meaningful participation in AI-mediated social life itself.