Vitalik Buterin suggests the use of AI ‘stewards’ to reform governance in DAOs.

The system would implement zero-knowledge proofs and secure environments (MPC/TEEs) to safeguard voter identities and confidential information while deterring coercion and bribery.

What to know:

• Buterin suggested the implementation of individual AI models influenced by users’ values to automate voting on numerous DAO decisions, tackling low engagement and the delegation of votes to significant token holders.
• The framework would incorporate zero-knowledge proofs and secure environments (MPC/TEEs) to maintain the confidentiality of voter identities and sensitive data, while mitigating coercion and bribery.
• Prediction markets would motivate the submission of high-quality proposals and eliminate spam, with AI agents identifying only essential matters for human evaluation, thus automating standard governance involvement.

Ethereum co-founder Vitalik Buterin has proposed a comprehensive technical revision of decentralized autonomous organizations (DAOs), advocating for the utilization of personal artificial intelligence agents to privately cast votes for users and enhance digital governance scalability.

This proposal, shared on the social media platform X a month after Buterin criticized DAOs for declining participation and increased power centralization, seeks to encourage users to refrain from delegating their votes to major token holders.

STORY CONTINUES BELOWDon’t miss another story.Subscribe to the The Protocol Newsletter today. See all newslettersSign me upBy signing up, you will receive emails about CoinDesk products and you agree to our terms & conditions and privacy policy.

Instead, users would utilize their own AI models, developed based on their previous communications and expressed values, to vote on the multitude of decisions faced by DAOs.

“There are many thousands of decisions to make, involving many domains of expertise, and most people don’t have the time or skill to be experts in even one, let alone all of them.” Buterin stated. “So what can we do? We use personal LLMs to solve the attention problem.”

The first priority is maintaining the privacy of content, ensuring that sensitive information stays confidential. AI agents would function within secure environments such as multi-party computation (MPC) or trusted execution environments (TEEs), allowing them to handle private data without disclosing it on the public blockchain.

The second aspect is ensuring participant anonymity. Buterin advocated for the application of zero-knowledge proofs (ZKPs), a cryptographic method that enables users to demonstrate their voting eligibility without revealing their wallet address or the details of their vote.

This safeguards against coercion, bribery, and the phenomenon of whale watching, where smaller voters align their choices with those of larger token holders.

These AI overseers would facilitate routine governance participation and identify only significant issues for human examination.

To eliminate low-quality or spammy proposals, a growing challenge as generative AI inundates open forums, Buterin proposes the establishment of prediction markets. In these markets, agents could wager on the probability of proposals being accepted.

Successful wagers would yield rewards, promoting valuable contributions while penalizing irrelevant noise.

Buterin also recommended privacy-preserving technologies such as multi-party computation and trusted execution environments, which would enable AI agents to evaluate sensitive information, like job applications or legal disputes, without revealing it on a public blockchain.

Read more: From 2016 hack to $150M Endowment: the DAO’s second act focuses on Ethereum security