Article 4 evidence for HR: role matrix and training records

Article 4 evidence becomes practical when it is organised around roles. HR is a good example. A recruiter, hiring manager, HR operations lead and compliance officer may all touch the same AI system, but they do not need the same training.

The recruiter reviews shortlists. The hiring manager interprets candidate recommendations. HR operations manages process settings. Compliance asks whether the evidence file is complete. A generic AI awareness course does not reflect these different responsibilities.

Start with a role matrix

A role matrix is a simple table that links people to AI exposure. For each HR role, record:

• which AI systems they use or influence;
• which decisions or process steps are affected;
• which risks they must understand;
• which learning modules apply;
• which scenario assessment proves competence;
• which refresher cadence applies.

The matrix should be short enough to maintain. If it becomes a 40-page spreadsheet, nobody will update it.

Four HR roles to map first

Recruiters need to understand AI-assisted screening, matching, candidate communication, bias indicators and override logging.

Hiring managers need to understand that AI output is decision support, not objective truth. They need scenarios on shortlist review, interview preparation and avoiding automation bias.

HR business partners need to understand workforce analytics, people risk signals, monitoring boundaries and escalation when AI touches employment relationships.

Legal, privacy and compliance need to understand the evidence file: classification, data governance, transparency, human oversight, training records and vendor documentation.

These roles can share a foundation module, but the proof should diverge after that.

What a training record should contain

For Article 4, a useful HR training record includes more than a completion date. Store: