A civilization-scale systems approach to preserving human continuity, agency and resilience while enabling beneficial human-AI cooperation.
HCA asks how a civilization containing humans, increasingly capable AI, robotics and autonomous infrastructure can remain resilient against catastrophic failure, misuse, concentration of power and common-cause vulnerabilities. Its central proposition is simple: human preservation should become architectural rather than merely instrumental.
Structure consequential capabilities so that no single AI, human actor, organization or common control domain has a credible unilateral path to civilization-scale catastrophe.
Use independent and diverse monitoring, verification, authorization and fail-safe mechanisms to reduce common-mode failure.
Maintain sufficient independent human capability to sustain essential functions and recover from prolonged loss or compromise of AI services.
Protect human continuity, autonomy, agency, diversity and future choice.
Model civilization-scale dependencies and second- and third-order consequences.
Use diverse monitors, auditors and human oversight to watch consequential AI behavior.
Compartmentalize capabilities, permissions and critical functions.
Detect, isolate, contain and recover; fail toward less authority.
Retain independent human capability, biological continuity and recovery capacity.
Increasing AI capability must never produce decreasing compartmentalization or decreasing human recovery capability.
Explore the Six-Layer ArchitectureRevision 0 establishes a preliminary technical and organizational baseline intended to be challenged, reviewed, tested and improved. HCA is now moving into early organizational formation and is seeking qualified professional stakeholders across the disciplines needed to mature the architecture.
