Concept guide

AI literacy describes what someone understands. AI fluency shows up in how they work.

The terms overlap, but they are not interchangeable. AI literacy concerns knowledge and informed awareness; applied AI fluency concerns how a person uses that understanding within a real task.

Neither concept is captured completely by one score. A sound assessment chooses evidence that matches the question and states the limits of what the result can establish.

A practical distinction

AI literacy

Knowledge of AI concepts, capabilities, limitations and responsible-use considerations. It can be explored through instruction, discussion, quizzes and self-report.

Applied AI fluency

The demonstrated ability to frame work, use evidence, collaborate with AI, verify outputs, exercise judgment and deliver a useful result in context.

A person may understand AI concepts yet struggle to apply them under ambiguity. Someone may also complete a familiar task effectively without demonstrating broad conceptual knowledge. The assessment method should reflect the question being asked.

What different methods make observable

Key points

Knowledge checks
Can examine recognition, recall and conceptual understanding, but usually do not show how someone handles a complete work assignment.
Self-assessments
Can capture confidence and perception, but self-report is not independent evidence of applied performance.
Conversational diagnostics
Can explore reasoning through dialogue, though the participant may not need to navigate sources and produce a finished workplace deliverable.
Realistic work simulations
Can create a bounded work sample connecting task framing, source use, AI interaction, verification, judgment and final delivery.

Choose the evidence that fits the decision

Use literacy-oriented methods when the question concerns conceptual understanding, responsible-use awareness or learning comprehension. Use work simulations when the question concerns how a person applies AI in a realistic task.

A combined program may use both. The results should remain distinct: concept knowledge is not proof of applied performance, and one observed task is not proof of universal capability.

Keep interpretation bounded

Simulation results depend on scenario design, task familiarity, accessibility, time pressure, supplied materials and the quality of captured evidence.

A report should connect interpretation to observable evidence, disclose missing signals and avoid claims about future job performance or automated employment decisions.

Discuss an assessment pilot

Tell us what kind of AI-enabled work you want to understand. We will help determine whether an existing assessment scenario fits the context.

Request early access