Panel Discussion: How to Build Reliable Global AI Governance without Creating a Monster