Ethereum co-founder Vitalik Buterin has proposed a novel solution to the chronic low participation in DAO governance: personalized AI voting agents. In a recent statement, he argued that current delegation models are disempowering and lead to centralization. His concept involves users training an AI model on their personal views and history, which would then vote autonomously on their behalf within privacy-preserving systems.
Vitalik Buterin has identified a critical flaw in DAO governance, noting that voting suffers from serious participation problems. He proposes that personal AI agents, trained on a user’s messages and past decisions, could vote on proposals exactly as the user would.
Buterin explained the core issue, stating, “Making good decisions often cannot come from a linear process of taking people’s views that are based only on their own information, and averaging them.” Most community members lack the time to evaluate thousands of complex decisions, which concentrates power with a few large token holders.
He has previously criticized the state of DAOs, as he noted last month regarding their low participation and growing centralization. Buterin stated that the common alternative, delegation, is fundamentally disempowering for users.
A key part of his AI agent proposal involves maintaining user privacy and security. These agents would operate inside secure systems like trusted execution environments to process sensitive data without public exposure.
To further protect voter identity, the use of zero-knowledge proofs has been suggested. This would allow users to prove voting eligibility without revealing their wallet address or specific decision, potentially reducing risks like bribery.
For practical implementation, AI agents could handle routine votes autonomously and alert users only for significant matters. To manage spam proposals, the idea of using prediction markets is on the table, where AI agents would financially back strong proposals.
