A public, citable framework for measuring AI proficiency.

This framework describes how Emeri measures AI proficiency. It is published in full so organizations, researchers, and educators can examine and reference it. Cite the published page when you refer to the framework.

What this framework assesses

The framework measures practical AI proficiency: how capably someone works with AI tools to complete real tasks. It does not assess theoretical knowledge, model internals, or vendor-specific features. The focus is on behaviors that decide whether AI makes work faster, better, and safe to trust.

Six competencies

AI proficiency breaks down into six practical skills. Each is graded independently, then blended into an overall band and percentile.

Asking well

Asking for what you need clearly enough to get useful results.

Checking the work

Spotting mistakes, checking sources, and knowing what to trust.

Putting it together

Turning prompts, tools, and steps into a repeatable workflow.

Getting real work done

Using AI to finish real work, not just make a first draft.

Giving good context

Giving AI the background, data, and limits it needs.

Keeping it safe

Protecting private information and using good judgment.

Four bands (E1 to E4)

Proficiency is measured on a four-band scale from foundational to exceptional. Each band describes observable behaviors, not just skill level. E1 is an earned credential — performance below E1 receives no grade.

Role calibration

Questions are tailored to the test-taker’s role (marketing, engineering, support, etc.) so scenarios feel relevant and the standard remains fair. A marketing professional and a software engineer see different tasks, but both are graded against the same behavioral rubrics. An E3 in marketing and an E3 in engineering both represent distinguished proficiency.

Open-AI assessment

Test-takers are encouraged to use AI during the assessment. Emeri grades the quality of finished work and the judgment behind it, not whether someone can do the task unaided. This reflects how AI is actually used: as a tool that amplifies capability when used well and introduces risk when used carelessly.

Grading method

Assessments combine multiple-choice questions (testing recognition of good and bad AI practices) with open-response scenarios (testing the ability to produce real work). Scenarios are graded by an AI examiner applying human-written rubrics. Those shared criteria are intended to keep each grading decision anchored to the published standard.

The full grading methodology, including pass bar and retake policy, is published on the Methodology page.

What this framework does not claim

This framework has not been psychometrically validated through external research. Emeri does not claim test-retest reliability statistics, external expert panel review, or accreditation from industry bodies. The bands and competencies describe practical professional behaviors and are graded against rubrics written by the Emeri team; they have not been validated through peer-reviewed studies.

The framework is practical and honest about its limits. It measures how someone works with AI on the day they were assessed, not their potential, domain expertise, or overall job performance.

Using this framework

Organizations can reference this framework when considering internal AI assessments, training programs, or proficiency targets for hiring. Educators and researchers can cite it when comparing approaches to AI capability assessment.

For a behavior-based reference, see the AI Skills Matrix, which maps the six competencies to observable foundational, developing, and stronger behaviors.

Citing this framework

Emeri. (2026). AI Skills Assessment Framework. Retrieved from https://emeri.org/resources/ai-skills-assessment-framework