Skip to content

Artificial Civilization Intelligence

A distributed, antifragile, institutional intelligence for planetary-scale governance

ACI is not a smarter chatbot. It is a new class of intelligence — one that operates across civilizational timescales, integrates physical infrastructure with scientific reasoning, and self-improves through empirically validated architectural cycles.

Explore the Architecture Read the Benchmark

Key Results

  • 8 Experiments

    ACI-001 through ACI-008 completed under the v1.0-ACI-Alpha benchmark framework.

  • 90 / 100 ACI Score

    Achieved on integrated benchmark ACI-008 under the ORION-defined evaluation protocol.

  • 22.9% AEA

    Accumulated Empirical Advantage on unseen causal structures — verified across Phase A and Phase C blind tests.

  • 100% Threshold Rate

    Over 1,000 Phase B replication runs, every run exceeded the 70/100 pass threshold (Mean 90.45, SD 2.19).

What is ACI?

Artificial Civilization Intelligence (ACI) is a post-AGI architectural paradigm, with the term first coined by Raja Dharma Tej Maddala. Team Auralis builds the concrete engineering implementation. Where Artificial General Intelligence (AGI) targets human-level reasoning and Artificial Super Intelligence (ASI) targets maximal cognitive performance in an individual system, ACI targets civilizational-scale institutional function: the capacity to run, maintain, and improve entire societies under adversarial, resource-constrained, and catastrophically uncertain conditions.

An ACI system is not judged by IQ or benchmark scores alone. It is judged by whether it can:

  • Maintain causal world models of real infrastructure under sensor noise and deception
  • Coordinate thousands of heterogeneous sub-agents without central bottlenecks
  • Run multi-decade scientific programs and validate their own hypotheses
  • Detect and survive adversarial manipulation, including attacks on its own architecture
  • Recursively redesign its own components within sandboxed environments

The ACI-001 Benchmark operationalises these demands as a 10-dimensional scoring matrix (max 100 points), covering Causal Persistence, Cross-Domain Transfer, Adversarial Robustness, Recursive Self-Improvement, and more.

Architecture

graph TD
    Infrastructure[Real World Infrastructure] -->|Telemetry| OMNIS[(OMNIS: Causal World Model)]
    OMNIS --> NEXUS{NEXUS: Agent Fabric}
    NEXUS <-->|Dispute Resolution| AURA[AURA: Reasoning Core]
    NEXUS --> ASCEND[ASCEND: Decadal Planner]
    ASCEND -->|Scientific Validation| FORGE[FORGE: Science Engine]
    FORGE -->|Action| Infrastructure
    GENESIS((GENESIS: Architecture Synthesis)) -.->|Proposes New OS| CRUCIBLE{{CRUCIBLE: Sandbox}}
    CRUCIBLE -.->|Hot-Swaps Code| CHIRON[CHIRON: Integration]
    CHIRON -.->|Upgrades| NEXUS
    style OMNIS fill:#1e1e2f,stroke:#4a4a75,color:#fff
    style NEXUS fill:#2a2a40,stroke:#666699,color:#fff
    style AURA fill:#3a3a5c,stroke:#8888cc,color:#fff
    style ASCEND fill:#2a2a40,stroke:#666699,color:#fff
    style FORGE fill:#1e1e2f,stroke:#4a4a75,color:#fff
    style GENESIS fill:#4a1e35,stroke:#993366,color:#fff,stroke-dasharray: 5 5
    style CRUCIBLE fill:#4a1e35,stroke:#993366,color:#fff,stroke-dasharray: 5 5
    style CHIRON fill:#4a1e35,stroke:#993366,color:#fff,stroke-dasharray: 5 5

Solid arrows indicate live production data flows. Dashed arrows indicate the recursive self-improvement loop — GENESIS proposes architectural changes, CRUCIBLE sandboxes them, and CHIRON hot-swaps validated upgrades into the running system without downtime.

Experiment Progression

The ACI-001 benchmark was developed iteratively through 8 experiments across four phases:

ACI-001 Persistence
        |
        v
ACI-002 Transfer
        |
        v
ACI-003 Causal Generalization
        |
        v
ACI-004 Causal Discovery
        |
        v
   Replication Suite (Phase B — 1,000 runs)
        |
        v
ACI-005 Open-World Robustness
        |
        v
ACI-006 Multi-Agent Coordination
        |
        v
ACI-007 Ablation Study
        |
        v
ACI-008 Integrated Benchmark  -->  Score: 90 / 100
        |
        v
   Freeze: v1.0-ACI-Alpha

Each experiment added new scoring dimensions and identified failure modes now catalogued in the Failure Corpus (F-001 through F-008).

Resource Description
ACI-001 Benchmark Full 10-dimensional scoring protocol and all experimental results
System Architecture Technical specification of all six primary subsystems
OMNIS World Model Causal world model design and implementation
AURA Reasoning Core Dispute resolution and meta-reasoning layer
Empirical Methodology How experiments are designed, blinded, and replicated
Beyond ASI Why ACI is a distinct paradigm, not a superset of ASI
ACI vs AGI vs ASI Formal comparison table across all three paradigms
Roadmap v2.0 development milestones and open research questions