Hiring & AI Glossary

Key terms in structured hiring, AI evaluation, and recruitment compliance.

These definitions describe review concepts and evidence context. They do not describe an automated hiring decision or product ranking.

Structured Interview

An interview method where candidates for the same role receive the same planned core questions in a consistent order and reviewers use a common rating approach. Relevant follow-up probes can be used to clarify evidence while preserving comparability.

Competency-Based Evaluation

A review approach that examines interview evidence against predefined, role-specific competencies — such as problem solving, communication, or technical depth — instead of relying on general impressions. Reviewers document the supporting evidence and any gaps before making a human decision.

Evidence-Linked Evaluation

A review method that connects an assessment to its supporting source, such as a relevant interview excerpt, and records the reviewer’s reasoning and unresolved gaps. The source can be checked and challenged by the people responsible for the hiring decision.

Adverse Impact

A situation where a selection practice disproportionately affects members of a protected group. In U.S. guidance, the four-fifths rule compares a group’s selection rate with the group that has the highest selection rate; a ratio below 80% can be a signal for further review, not a conclusive legal test. Also called disparate impact.

STAR Method

A structured framework for answering behavioral interview questions. STAR stands for Situation (context), Task (responsibility), Action (what the candidate did), and Result (outcome and impact). Interviewers use STAR to assess the quality and completeness of candidate responses.

Selection Validity

The extent to which evidence from a selection method is associated with the job outcome the method is intended to assess. Validity depends on the role, criterion, study design, population and how the method is implemented.

Deterministic AI

A system design intended to produce repeatable outputs under specified conditions. Settings such as a low temperature or fixed seed can reduce variation, but do not by themselves guarantee identical results across model, provider or software changes. Versioning and repeatability tests remain necessary.

Human-in-the-Loop

A system design where accountable people can understand, review, challenge and override AI-assisted output before it influences an outcome. Meaningful oversight requires suitable authority, information, time and competence; the applicable legal duties depend on the system and its use.

High-Risk AI System

A regulatory classification used by the EU AI Act for AI systems that meet specified conditions. Annex III lists certain intended uses in employment, worker management and access to self-employment. Classification depends on the system’s intended purpose and context, so a specific deployment needs legal and technical assessment.

DSAR (Data Subject Access Request)

A request through which a person exercises the GDPR right of access to their personal data and related information. Access and erasure are separate rights with different conditions and exceptions. Recruitment records in scope depend on what the organization processes and the applicable retention and legal obligations.

Time-to-Hire

A recruiting cycle-time metric, commonly measured from a candidate entering the hiring process to accepting an offer. Teams should document their start and end events because definitions vary across organizations and reporting systems.

Interviewer Calibration

The process of aligning interviewers on role criteria, observable evidence and how to document their assessments. Reviewers discuss differences and missing information while retaining independent judgment and a clear record of their reasoning.

Behavioral Anchors

Concrete descriptions of observable evidence associated with levels on a role-specific rating scale. Anchors give reviewers a shared reference, but still require evidence, calibration and accountable judgment.

AI Bar Raiser

A product-style label sometimes used for AI-assisted interviewer guidance. It is not a standardized professional or legal term. Capabilities, evidence quality and human-control boundaries depend on the specific tool and configuration.

Talent Pool

A collection of prospective candidates who may be considered for current or future roles. How people enter, remain in or are matched from a talent pool depends on the organization’s process, lawful basis, retention rules and the tools in use.