People create the knowledge, examples, judgments, and communities that make AI useful. They should also have meaningful opportunities to shape its goals and share the value of their contributions. That is the central idea behind our proposed Decentralized Engagement Organization, or DEO.
In this series, DEO names an organizational design, not an established scientific solution or a standardized legal structure. Its purpose is to make participation an operating part of AI alignment.
Six steps that connect contribution to change
- Listen: recruit affected people, including those who rarely participate.
- Deliberate: compare proposals, evidence, and tradeoffs.
- Decide: document who has authority and which protections cannot be overridden.
- Test: evaluate the resulting AI behavior on agreed cases.
- Reward: compensate defined contributions from an identified budget.
- Revise: publish failures and reopen decisions when evidence changes.
What makes it decentralized?
Meaningful authority must exist beyond one operator. A pilot could give a rotating participant council control over a small review budget, let members elect reviewers, and require organizers to answer appeals. Publish the limits of that authority. An advisory survey controlled entirely by a founder should not be presented as decentralized governance.
Blockchain may help record decisions or administer a treasury in some implementations. It is optional. It does not establish that the input is truthful, that participants are independent, or that a decision is fair.
Voice should not be for sale
Keep participation rights separate from wealth. Buying more tokens should not automatically mean more influence over protections affecting everyone. Reward time, verified findings, careful reviews, and useful resources. Do not reward signatures or agreement with the organizer.
Use a base stipend where possible, transparent review criteria, conflict disclosures, and an appeal process. Preserve criticism even when it challenges the project itself.
What would count as success?
A DEO succeeds when it demonstrably improves decisions and outcomes: participants can identify changes they influenced; reviewers find meaningful errors; benefits reach contributors; and people can challenge authority without losing standing. Compare these results against a simpler process. If a conventional review panel works better, learn from that.
Our position is that DEOs deserve serious development as part of alignment. Their success must be demonstrated, and technical safety work remains necessary.
Run the 30-day pilot. Related perspective: platform cooperativism.
