Reference

Hiring assessment glossary

The concepts behind AI-generated, quality-gated hiring assessments — defined plainly, each linked to a deeper explainer.

4D framework: A four-part model for assessing AI fluency: Delegation (deciding what to hand to AI), Description (specifying the task clearly), Discernment (evaluating what comes back), and Diligence (owning the result). Weighted per role in the AI Fluency pillar.
Adverse impact: A hiring practice that disproportionately screens out a protected group, measured by the four-fifths rule. It can occur even without intent, which is why assessments should be audited for it.
AI fluency: A measure of whether a candidate works effectively and responsibly with AI tools — knowing when to trust output, when to verify, and how to use models with judgement.
AI Sandbox: A live exercise where a candidate uses AI tools to fix or build a working artifact, measuring prompt craft, verification instinct and judgement — the applied version of AI fluency.
BARS (behaviourally-anchored rating scale): A scoring rubric that defines each level by specific, observable behaviours rather than vague adjectives, so two reviewers rate the same answer the same way.
Behavioural pillar: The hiring pillar that scores work style and traits against behaviourally-anchored rubrics instead of gut feel, turning soft signals into comparable ones.
Cognitive pillar: The hiring pillar measuring reasoning, pattern-finding and learning speed — a strong predictor of potential, but weak as a standalone gate for job skill.
Concept tree: A structured map of the sub-skills a competency contains. Questions are generated per node so an assessment covers a skill systematically instead of sampling it at random.
Domain pillar: The hiring pillar that tests whether a candidate can actually do the work — the role's real skills — using per-role questions grounded in concept trees.
Five pillars: The five distinct, separately-scored signals of a hire: cognitive ability, situational judgement, behavioural traits, domain skill and AI fluency, weighted per role.
Integrity engine: Layered integrity signals combined into one composite risk score per candidate, with recruiter-visible flags.
Per-job generation: Building an assessment on demand from a job description rather than selecting it from a shared library — so there is no static, shared answer key to leak.
Quality gate: A two-phase check — structural rules plus an independent AI judge — that every generated question passes before a candidate can see it, with failures quarantined.
Quality of hire: A feedback loop that validates assessment predictions against real post-hire outcomes, so the assessment learns whether it actually predicted performance.
Situational judgement: The hiring pillar that presents realistic, ambiguous work scenarios and scores how a candidate prioritises, escalates and decides — judgement rather than recall.
Three-tier bank: A question-bank architecture that keeps content fresh per job and deduplicates by content fingerprint, so answers cannot be memorised or leaked.

See your first AI-generated assessment in 5 minutes

No credit card. No question library to learn. Paste a job description and watch H-Evaluate build.

Request early access