From Particularity to Collective Agency
People struggle to decide together across real difference: of faith, politics, culture, history and power. Artificial intelligence has made the struggle urgent, adding shared questions about what AI may do, what stays human, and who decides. Decisions made quickly tend to erase real differences and silence the people with the least power; decisions handed down tend to lack legitimacy with the people who must live by them. AI-mediated deliberation can help, but it introduces a new risk: an AI summary can represent people accurately while leaving them with no real influence over what is decided. This paper describes a general method for moving from individual judgment to shared decisions, building on the Personal Moral Compass (Edmonds-Allen 2026). Each person first forms and approves their own account of what matters, then chooses what to contribute. AI organizes contributions, drafts, and checks fidelity; people correct how they are represented, decide through their own legitimate procedures, and record each decision with its reasoning, dissent, and review date. Decisions are sorted by how much consistency they need, and on shared work the strictest stated boundary holds. Two implementations are built and running in higher education: the Department Compass, for academic departments setting AI guidelines and syllabus standards, and the Group Project Compass, for student teams agreeing how AI will and won't be used. The paper sets out the method, both implementations, five proposed applications (community governance that includes people at the margins; dialogue across deep difference; fair representation of cultures and identities; verifiable human control of AI; and leadership development and credentialing), safeguards and limitations, and a plan for evaluation. It distinguishes four dimensions of participation (expression, recognition, influence, and authority) and proposes that the method be judged on all four, not on agreement alone. No outcome claims are made.
Authors
- Marian Edmonds-Allen
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-14
- DOI
- https://doi.org/10.5281/zenodo.24234850
- Primary Topic
- Ethics and Social Impacts of AI
- Type
- article
- Field-Weighted Citation Impact
- 0.00