AGI-only artificial-life research program

GHOST

Can a living ecology make competence and individuality emerge?

GHOST tests whether a digital organism can earn adaptive behavior through survival, reproduction, heredity, and ecological selection rather than a ranked external fitness grader.

Bounded scientific core certified · OLK-v4 active
Open source record in Notion
ANIMUS MACHINAE / PROJECT RECORD
Latest completed directiveD-211
CertifiedBounded Levels 1–4 core
Current frontierOLK closed-loop structural efficacy
Not establishedScalable brain or operational AGI
01

What has been shown

Living ecology increased inherited competence without ranked reproductive grading. Different lifetime histories produced persistent adaptive individuality. Learning-specific plasticity parameters were heritable and causally selectable. Ecological selection enriched the inherited capacity to become an adaptive individual.

02

Why the boundary matters

A bounded smoke or local structural effect is not AGI. GHOST has not yet established scalable searchability, broad continual learning, transfer, abstraction, long-horizon agency, open-ended capability growth, or ecological re-earning by a scalable brain.

03

How the research is governed

The project moves only when a frozen gate passes, false explanations are excluded, and the required independent audit certifies the claim. Failed architectures stay in the record instead of being quietly rewritten as progress.

Claim boundary

GHOST is a scientific program, not a companion, avatar, product deployment, or claim of operational AGI.

Read the source record in Notion