Caneni defined term · not a discovery flagship

AI Oversight Evidence

Not AI watching humans. Caneni uses “AI Oversight Evidence” to mean evidence of human oversight around AI-assisted decisions — a record that a person exercised judgment, not that an AI supervised people.

Because the phrase “AI oversight” is read two ways, Caneni's current public professional term is Human Oversight Evidence.

Canon Atlas holds the deeper definitions. This page keeps the public boundary language readable and scoped.

Two Meanings of Evidence

Research meaning

Evidence about AI oversight: empirical studies of how AI monitoring changes human behavior, errors, incentives, or automation bias.

Caneni meaning

A structured, attributable record of a person's stated considerations, reported opportunities to intervene, and declared responsibilities in an AI-assisted decision, alongside information about AI contributions with its source identified.

Provenance: AI Oversight Evidence is a historical Caneni term / C-AOE notation. Current public professional term: Human Oversight Evidence.

For the professional application and Article 14 context → Human Oversight Evidence

What It Records

Context seen. What information the human had before acting.

AI boundary. Where AI contributed and where human judgment had to take over.

Consequence held. What outcome or risk the person accepted as theirs to hold.

Uncertainty acknowledged. What was not known, not reliable, or not yet settled.

Return condition. When the decision should be revisited if facts or consequences change.

Disclosure Boundary

Caneni does not infer attention, understanding, judgment quality, or oversight adequacy from behavioural telemetry. The public distinction is between attributable human action and automatic system action, not between conscious and unconscious mental states. Integrity claims are scoped: a record can be made reviewable and tamper-evident for a stated environment, not impossible to forge in every context.

Why Logs Are Not Enough

An AI audit log can show inputs, outputs, confidence scores, and actions. A workflow log can show that someone clicked approve.

A click records a transaction. AI Oversight Evidence supports examination of the human judgment behind it.

This is why Caneni pairs AI governance language with reviewable human judgment and the consequence record.

Article 14 Context

EU AI Act Article 14 is a relevant use context because it concerns human oversight for high-risk AI systems. Caneni does not claim that Article 14 requires Caneni, or that a Caneni record automatically satisfies any legal duty.

Caneni frames the evidence layer: how human oversight can be made more visible, reviewable, and attributable when an AI-assisted decision is later questioned.

Read Article 14 framing Protect for boards

What This Is Not

≠ Proof that oversight was effective ≠ Proof that the decision was correct ≠ Compliance certification ≠ Audit report ≠ Legal opinion ≠ Claim that Article 14 requires Caneni ≠ A rating or score

Human Oversight Evidence DAHS Ask Apotrop