A multilevel framework for AI governance

Ethics alone isn't enough: The paper emphasizes that while many organizations have created ethical guidelines for AI, these principles won't actually be implemented without a formal governance structure in place—highlighting a critical gap between stated values and real-world practice.

Three stakeholders must work together: Effective AI governance requires coordination among governments, corporations, and citizens as interdependent partners, rather than any single group acting alone to manage AI risks and benefits.

Trust requires multiple dimensions: The framework identifies three key trust components—competence, integrity, and benevolence—suggesting that people need to believe AI systems are technically capable, honestly designed, and genuinely intended to help them.

Governance structure enables policy: By combining multilevel governance with trust dimensions, the paper provides a practical foundation that can guide both how companies design AI systems and how governments should regulate them.

User experience and public policy are connected: The framework suggests that improving how people actually experience and interact with AI systems should directly inform the development of public policies governing AI, creating a feedback loop between practice and regulation.