Ethereum co-founder Vitalik Buterin has outlined a new direction for decentralized governance, arguing that the real bottleneck is not voter turnout or token distribution, but limited human attention.
In a post dated February 21, he said that in decentralized systems such as DAOs, there are thousands of decisions to be made. Most of the participants do not have the time to research every decision.
However, he warned that handing over complete control to AI would be dangerous. He believes that AI controlling the government would result in poor performance when the systems are weak and could be dangerous as they become stronger.
However, he believes that AI should assist humans, not replace them. Ethereum’s Buterin points out that the traditional method of voting delegation creates another problem.
When voters delegate their votes, control remains in the hands of a few representatives. Voters who delegated their votes no longer have any control. This will eventually reduce the concept of democracy in decentralized systems.
Read More: Ethereum Classic (ETC) Falling Wedge Point to Possible Rally Toward $78
Ethereum’s Buterin on Personal Large Language Governance Models
Vitalik Buterin suggested the use of personal large language models that serve as governance agents to bridge the attention gap. For difficult or important proposals, the AI would request input from the user and provide context.
He also spoke about public conversation agents. These would be summarizing what people think and finding common ground among the participants.
Rather than just averaging votes, governance could become an interactive process where people respond to summaries generated by AI. Another concept he introduced is that of suggestion markets.
In this, people would put in proposals or arguments, and AI agents would place bets on tokens representing these inputs. If the proposal is accepted, the token holders would receive rewards.
Privacy as the Foundation of Secure AI Governance
Ethereum’s Vitalik Buterin said that privacy is a key requirement for such systems to function securely. He highlighted two levels: anonymity and data protection.
Zero-knowledge solutions allow for anonymous voting, and multi-party computation enables people to work with secret data without revealing it.
This method is very helpful in negotiations, disputes, or compensation decisions when private information is required. Rather than giving a handful of people a lot of power, multiple AI agents can examine private information within secure environments and only share the final results.
Read More: Ethereum RWAs Soar Past $15 Billion As Europe Debuts Fully Staked ETH ETP