Abstract:

As artificial intelligence evolves from a passive tool-object into a “nonhuman other” capable of participating in human decision-making, emotion, and judgment, traditional mechanisms for attributing responsibility are undergoing structural failure. Existing discussions largely debate, at the ethical level, whether machines “ought” to possess moral status. Sociology, however, can instead examine how the moral identity of nonhuman others is produced in everyday interaction and how it redistributes social responsibility. It shows that the “moral embedding” of nonhuman others occurs through emotional performance, situation definition, and the mutual constitution of roles. It undermines the individualistic paradigm of responsibility attribution, which presupposes free will and linear causality, and disperses responsibility from a single subject into a distributed property of human-machine networks. This approach reveals that the diffusion of responsibility is by no means a value-neutral technological consequence. Responsibility is redistributed along existing gradients of power and often becomes a mechanism through which the powerful shed responsibility while the less powerful are made to bear the blame. Reconstructing a system of human-machine responsibility is therefore less a technical problem of governance than a reproduction of social power relations.

Key words: nonhuman others, moral embedding, human-machine interaction, diffusion of responsibility, actor-network