Virtue and Artificial Intelligence

Abstract

Two questions concerning artificial intelligence (AI) and virtue are explored. The first question is whether AI is capable of virtue, and the second question explores if AI can assist humans in the acquisition of virtue. It is argued that AI cannot be a moral agent and therefore cannot genuinely be virtuous. However, AI can perform behaviors that are in accordance with virtuous behavior. The notion of "virtue-by-proxy" is introduced in which virtuous programmers can strive to design AI programs that are trained to mimic certain virtues or behave in accordance with virtue. Next, it is argued that since AI systems can nudge humans toward repeated practices and habits, they will inevitably shape and form users. Thus, AI systems might be designed to assist humans with virtue formation; likewise, they could be misdirected to encourage certain vices. Finally, the concept of virtue in the Christian tradition is contrasted with secular notions of virtue and is used to inform limits on the role of AI in virtue formation.