Definition
Percentage of AI-assigned work packages for regulated activities where the assigned worker's NRC-mandated qualifications, license status, and certification currency were verified at or before assignment time.
The Measurement Gap
When AI systems assign workers to maintenance activities on regulated nuclear equipment, the assignment must account for NRC-mandated qualifications that go beyond standard skill codes: licensed operator status under 10 CFR 55, training requirements under 10 CFR 50.120, radiation worker certifications under 10 CFR 20, and ASME/QA certifications. Legacy workforce metrics track training completion rates and overall qualification percentages. Human performance error rate tracks execution mistakes. Neither measures whether the AI assignment system verified qualification currency at the moment of assignment. The invisible failure mode is that the AI assigns based on skill codes and availability while assuming qualification compliance, but qualification databases may be stale, credentials may have expired between updates, or the AI may not distinguish between a worker qualified for general maintenance and one qualified for the specific regulated activity. No existing metric detects the gap between "assigned" and "verified qualified at assignment time."
Decision Supported
Is the AI work assignment system checking that assigned personnel are regulatorily qualified for the specific task, or is it assigning based on skill codes and availability while assuming qualification compliance?
Formula
Plain language: Count AI-assigned regulated work packages. For each, check whether the assigned worker's regulatory qualifications (NRC license, rad worker certification, QA certs, procedure qualifications) were verified as current at the time of assignment. Divide verified by total.
Notation:
RQVR = Verified_regulated_assignments / Total_regulated_assignments
Variables:
- Verified_regulated_assignments = AI-assigned work packages for regulated activities where all required regulatory qualifications were confirmed current at assignment time (count)
- Total_regulated_assignments = all AI-assigned work packages for activities requiring regulatory qualifications (count)
- Regulatory qualifications include: RO/SRO license status and medical currency, radiation worker qualification tier, ASME Section XI qualification (VT-1, VT-2, VT-3, UT, MT, PT, RT), Appendix B QA procedure qualifications, 10 CFR 50.120 required training currency
Sign convention: 1.0 = every regulated assignment was verified. Lower is worse. In nuclear, this metric has a zero-defect expectation: any gap is a potential NRC finding.
Aggregation period: Weekly (fast enough to catch problems before they accumulate).
Edge Cases
- Qualification expired between assignment and execution: track separately as "assignment-to-execution qualification decay". The AI verified at assignment but the credential expired before work started
- Worker holds qualification but it's in a different plant's system (fleet qualification transfer): count as unverified unless the receiving plant's qualification system confirms it
- AI assigns a crew rather than an individual: all crew members on regulated activities must be verified. Count as unverified if any crew member lacks the required qualification
Proposed formula, pending validation pilots.