AI virtues -- The missing link in putting AI ethics into practice

Principles alone aren't enough: While major AI ethics initiatives have established important principles and standards, the paper finds these have largely failed to translate into actual practice—a gap that simply adding more practical guidelines hasn't solved.

AI virtues provide missing motivation: The paper proposes that ethical AI development requires cultivating four core "virtues"—justice, honesty, responsibility, and care—that represent the actual motivational mindsets needed for people to make ethical decisions, not just follow rules.

Psychology matters more than we acknowledge: Two additional virtues, prudence and fortitude, are needed to help AI developers overcome "bounded ethicality"—the hidden psychological forces and biases that prevent people from making ethical choices even when they want to, which current AI ethics frameworks largely ignore.

Organizations must actively cultivate virtue: The paper argues that responsible AI requires deliberate organizational measures to develop these virtues in teams, rather than assuming ethical behavior will follow from publishing principles or policies.

A character-based approach is needed: By drawing on virtue ethics (focusing on who people are and what motivates them), rather than just rules or outcomes, the paper suggests a fundamentally different path to making AI ethics real in practice.