Measurement methodology
How AISkillProof measures AI capability
AISkillProof bases results on observable work behavior: what someone creates, checks, decides or structures in realistic AI tasks. A score is only useful when its evidence, rubric and version are traceable.
Last updated: 16 September 2026
Six stable capabilities
The current skill layer measures AI Literacy, Prompting & Instruction, Tool Mastery, AI Judgment, Workflow & Automation and AI Governance. Assessments and learning use the same skill keys so evidence remains comparable over time.
Evidence from real tasks
A measurement can combine objective scenarios, prompts, practical tasks, ranking, matching, completion and other work samples where appropriate. Candidates must create, check or decide; self-rating is not core evidence.
Explainable scoring
Questions with a fixed answer key are scored automatically on the server: scenarios, multiple choice, ranking, matching and completion questions with accepted answers. Free text and practical responses do not earn skill points. Scoring uses no AI or external grading service.
Versions and audit trail
Attempts retain assessment, scoring and rubric versions. Skill evidence and reviews are linked to the concrete attempt and grading source, allowing the reason for a score to be reconstructed later.
Readiness without pretend benchmarks
AI Readiness is built only from actually scored capability evidence. Missing skills remain visibly under-measured. AISkillProof does not show an external percentile or Belgian benchmark until cohort size, comparability and privacy make that defensible.
Reassess with fresh evidence
After targeted learning, the same capability is measured again with a different but comparable item set. Uplift compares the skill score on the linked baseline with that same skill on reassessment, not two arbitrary total test scores.
What the score does and does not assess
Scores use answers that can be checked against a fixed answer key. Length, formatting and keywords in free text earn no points. Reports show the assessed share of the assessment; this percentage does not measure correctness. Free text remains available for practice and feedback, but its content quality is not automatically assessed.
What a score does not mean
An AISkillProof score is evidence within a specific skill and assessment version. It is not a universal predictor of job performance, not an automated HR decision and not an officially recognized certificate unless that is explicitly established for a separate credential.