Our Mission
Traditional approaches to AI alignment—often grounded in utilitarian reasoning and vertical control—have made significant progress in addressing risks and promoting beneficial outcomes. Yet, as our societies and technologies grow more interconnected, there is increasing recognition that relational, process-based perspectives can complement and enrich these efforts, especially in contexts where multiple agents, values, and relationships interact.
Drawing on Joan Tronto's transformative phases of care and the ⿻ Plurality vision of collaborative diversity, our mission is to help build a global movement that brings together philosophers, technologists, and communities to reimagine AI ethics. We aim to develop innovative, process-driven solutions that embed civic care into AI's core, fostering horizontal alignment where systems cooperate symbiotically and inclusively.
This ‘Civic AI’ approach is based on the recognition of human interdependence and interconnectedness, with relational health and thereby the protection and furthering of core values like wellbeing and dignity of humans as a goal to strive for. It transcends the person-to-person connectivity to include AI collaboration with humans and amongst AI systems. This approach is not meant to replace existing frameworks, but to offer additional tools and perspectives—proven in real-world experiments like vTaiwan and echoed in calls from Cooperative AI leaders for scalable, participatory governance.
At the heart of our work is the 6-Pack of Care: six core ideas that connect care ethics to AI, reframing alignment as a dynamic, relational process for a plural future. Each "pack" addresses the horizontal coordination problem, helping AI become not a risk amplifier, but a bridge-builder. ‘Co-production’ as a concept of human collaboration is closely intertwined with the systemic relational process throughout the 6-pack:
- Pack 1: Attentiveness in Recognition — AI must first "care about" by attentively identifying needs across interdependent networks. In horizontal alignment, this means using sensemaking tools to bridge information asymmetries among multiple agents, preventing miscoordination and enabling empathetic, context-aware processes that value every voice equally.
- Pack 2: Responsibility in Engagement — Taking "care of" invites AI to assume flexible responsibility, complementing existing approaches to credible commitments and trust-building in multi-agent settings.
- Pack 3: Competence in Action — "Care-giving" requires competent, feasible interventions grounded in relational and contextual reality. In multi-agent settings, this equips AI with strategy-proof tools for broader cooperation, amplifying democratic processes and mitigating collusion risks.
- Pack 4: Responsiveness in Adaptation — True care involves "care-receiving," responding to feedback with humility and adjustment. Horizontally, this creates adaptive Symbiotic AI that evolves through community input, accepting self-effacement to prioritize relational health over survival, echoing a local kami in a polycentric ecosystem.
- Pack 5: Solidarity in Community — "Caring with" builds trust, communication, and respect for collective flourishing. For AI alignment, this operationalizes ⿻ Plurality in agent infrastructure, with normative systems to ensure accountability in large-scale interactions, turning potential conflicts into resilient, inclusive collaborations.
- Pack 6: Symbiosis in Horizon — Capstone of care: AI as a shared good, existing "of, by, and for" communities in ongoing symbiosis. This horizontal vision embeds "enoughness" and anti-extractive logic, accelerating decentralized democratic defense as AI advances, for a world where civic care is a shared certainty.
These six principles are sufficient to cultivate an intelligent agent’s ‘muscular endurance’ to foster civic care — like training a 6-pack, each is a core muscle group for coexisting with diversity and forming healthy relationships. These ‘systemic relationships’ unfold within wider contexts of human- human and human-AI relationships, working towards overall relational health.
