Buterin recently noted that DAO voting poses a serious participation problem, and its solution seems straight from the future. Instead of users manually voting on each proposal, he proposes letting personal AI agents do the work.
The idea is that each user would have their own AI model, trained on their past messages, opinions and decisions. This AI would then vote on the DAO proposals exactly as the user would.
“Often, making good decisions cannot come from a linear process of taking people’s views, based solely on their own information, and averaging them (even quadratically).”
Most people don’t have the time or expertise to vote on thousands of complex decisions. As a result, voting power often ends up being concentrated in the hands of a few large token holders. Buterin said:
“The usual solution, delegation, is disempowering”
What’s interesting is that this comes a month after Buterin took aim at DAOs for their low participation and increasing centralization.
A key part of the idea…
…ensure that users do not have to give away their privacy for convenience.
These AI agents would operate in secure systems, such as trusted execution environments and multi-party computations. This would allow them to process sensitive information without exposing it publicly.
Source:
To protect voters’ identities, Buterin suggested using zero-knowledge evidence.
This would allow users to prove they are eligible to vote without revealing their wallet address or their decision. This could also reduce risks such as corruption, coercion, or blindly copying large token holders.
AI agents could also automatically handle routine votes and alert users only when something important happens.
To deal with spammy proposals, prediction markets are the idea on the table. This is where AI agents would support strong propositions and ignore weak ones.
Final Summary
- Vitalik Buterin’s AI voting proposal could change DAO governance forever.
- If widely adopted, AI agents could make DAOs more scalable and truly decentralized.
