Vitalik Links AI Safety to Governance Theory
Ethereum co-founder Vitalik Buterin (
@VitalikButerin) says adversarial governance theory could offer new tools for AI safety.
He argues governance and AI safety share a similar principal agent problem.
In both cases, less capable overseers must control more sophisticated agents.
According to Buterin, mechanisms of design can be helpful in limiting exploitation by AI of human rules.
Furthermore, he emphasizes the possibility of collusion for both governance and AI systems.