learn

Etherom Learning System · Learning AX v0.1 · sealed

The Confirmation Model — learning is a confirmed change in capability, not content, completion, or recall. Two orthogonal axes, and a governed gate no AI worker can self-fulfil.

Endpoints

GET/api/learn— manifest
POST/api/learn/grasp— grasp level (L0–L9) + conf(G)
POST/api/learn/gate— Promote(x→Canonical): promote / hold / rollback

The two axes

Grasp — how well the learner demonstrates:

L0L3 explainL5 apply L6 transferL7 taughtL8 correctedL9 integrated

Truth — how trustworthy the claim:

UnresolvedObservedInterpreted SupportedConfirmed

The governed gate

Promote(x→Canonical) ⟺ (G≥L6) ∧ (T=Confirmed) ∧ (EvidenceScore≥τ_gov) ∧ HumanApproval ∧ (¬∃ contradiction)

HumanApproval is caller-supplied — the engine cannot self-fulfil it. That is the mechanical form of "no silent canonicalization."