Early Model

AI Capability Levels

A simple progression from using AI occasionally to shaping reliable human–AI systems.

Level 1 — Aware

Understands what AI can and cannot do. Uses basic tools with caution.

LiteracySafety

Level 2 — Interactive

Can ask useful questions, add context, and refine answers through dialogue.

PromptingContext

Level 3 — Critical

Checks assumptions, verifies claims, compares alternatives, and recognises weak output.

VerificationJudgement

Level 4 — Collaborative

Uses AI for sustained thinking, learning, creation, and problem solving across multiple steps.

ReflectionCo-creation

Level 5 — Systematic

Builds repeatable workflows, personal knowledge systems, and dependable AI-supported processes.

WorkflowKnowledge

Level 6 — Steward

Helps others use and train AI responsibly, improves standards, and governs complex AI use.

LeadershipGovernance

What the levels are not

They are not a ranking of human worth, academic intelligence, age, or professional status. A person may be highly advanced in one dimension and still developing in another.

The levels are intended as a language for development and discussion. Assessment methods can be designed later and should rely on authentic tasks, transparent criteria, and evidence of judgement.