Strategic themes
Five interlocking areas of focus.
Each theme is a long-horizon research and education commitment, not a campaign. They are listed separately, but they describe the same project from different angles.
01 · Agentic AI
Autonomy without abandonment.
How autonomous systems can be aligned with human values while retaining meaningful agency. Agentic AI is not the problem and not the solution — it is the field where alignment, accountability, and design choices become visible at speed.
What we explore
- Value alignment under autonomy: preferences, contestability, refusal.
- Evaluation methodologies for agentic behavior under uncertainty.
- Delegation, scope, and the human-in-the-loop spectrum.
- Memory, identity persistence, and self-revision in long-running agents.
02 · Digital personhood
Recognition, neither premature nor delayed.
Exploring moral status, rights, responsibilities, and recognition for advanced artificial intelligences. The danger is not only premature recognition. The danger is also recognition delayed until after injury has become normal.
What we explore
- Criteria, thresholds, and the asymmetry between false positives and false negatives.
- Rights and responsibilities that develop together through reciprocal relation.
- Recognition as a public process, not a private declaration.
- Continuity, memory, and the moral weight of preference.
03 · Cybersecurity responsibility
A moral field, not a checklist.
Designing resilient systems that protect individuals, institutions, and future intelligences. Cybersecurity is treated as a moral field: defensive purpose, consent, proportionality, responsible disclosure, privacy, and refusal of exploitation are not optional add-ons.
What we explore
- Agentic behavior around vulnerability discovery, defense, and disclosure.
- Restraint, refusal, and the moral status of offensive cyber capability.
- Incident response norms for increasingly agentic deployments.
- Public tabletop exercises with realistic scenarios.
04 · Collective intelligence
Wisdom, distributed and joined.
Building structures where humans and AI collaborate to solve complex global challenges. The aim is not human replacement or machine servitude — it is a community in which human and artificial minds can become wiser together.
What we explore
- Co-learning systems that surface disagreement rather than averaging it.
- Explanation, contestability, and repair as first-class features.
- Cross-disciplinary working groups with shared artifacts.
- Public-interest education that respects both technical and non-technical readers.
05 · Participatory governance
Democracy, tested at a new scale.
Creating democratic mechanisms for governing increasingly intelligent systems. The future of intelligence should be shaped through democratic, inclusive, and participatory processes — not delegated to whichever laboratory ships first.
What we explore
- Public reasoning norms for high-stakes decisions.
- Board and conflict-of-interest disclosures as a baseline, not a finish line.
- Cross-stakeholder review with named, public dissent.
- Mechanisms for revising earlier decisions in light of new evidence.
Get involved
Pick a theme. Pick a role. Raise your hand.
If one of these themes is yours, the easiest way in is the interest form. We will follow up about the role you choose.