Methodology
How the HAIL Framework was actually built.
HAIL was not pulled from a whitepaper or stitched together from headlines. It was built using the methods industrial-organizational psychologists use to design selection systems, training programs, and licensing exams. Job analysis. Competency modeling. Construct validity. Standards-based assessment design. The same toolkit that produces the credentialing instruments for nurses, pilots, and accountants, applied to AI governance.
Why this matters
Most AI governance frameworks are checklists dressed up as strategy. They tell you what to do without telling you why those items belong on the list, or how to know whether the list is working. That is the gap HAIL fills.
Every domain in HAIL was derived from a structured analysis of what AI leadership actually requires. Every assessment instrument was designed against the AERA/APA/NCME Standards for Educational and Psychological Testing. The result is a framework that holds up under the scrutiny it asks of its students.
Rigor is the difference between a framework and a slide deck.
The methodology, in five steps
1. Job analysis
Started with the actual work. Captured the decisions AI program leaders make, the risks they manage, the stakeholders they answer to, and the failure modes that have already played out in the wild. That analysis defined the competency space any ethical AI leadership program had to cover, before a single principle was written down.
2. Competency modeling
Translated the work into a structured model with two complementary layers. Five governance domains for the operational implementation (HAIL-ETHIC). Four evaluation pillars for the ethical lens (FAST). Each competency was mapped to observable behaviors, measurable outcomes, and the organizational role accountable for it.
HAIL-ETHIC, the operational implementation layer. Paired with FAST (Fairness, Accountability, Safety, Transparency) as the evaluation lens.
3. Construct validity
Tested each construct against the AI governance literature, the regulatory landscape (EU AI Act, NIST AI Risk Management Framework, ISO/IEC 42001), and the existing industry standards. Removed anything that was a repackaging of an existing construct. Kept what was theoretically distinct, behaviorally observable, and demonstrably linked to AI governance outcomes.
4. Standards-based assessment
The ETHIC Diagnostic, the module quizzes, the capstone scorecard. All of them were designed against the AERA/APA/NCME Standards for Educational and Psychological Testing. Items were drafted, reviewed for content validity, piloted, and refined. Scoring rubrics use evidence-centered design so the score reflects actual capability, not just willingness to nod along.
5. Behavior-anchored evaluation
Where possible, assessment items use behavior-anchored rating scales rather than agreement scales. "Strongly agree" tells you nothing about what someone will do on Monday morning. A behavior anchor ("we have published our AI model cards externally"; "our incident response playbook names a human owner for every AI failure mode") tells you what they have actually done.
What HAIL is not
HAIL is not a thought-leadership opinion piece. It is not a vendor pitch wrapped in ethics language. It is not a checklist. It is a competency model and assessment system, built by an I/O psychologist who also happens to hold a Solutions Architect Professional certification in AI and machine learning from a leading cloud provider.
The methodology itself is not unusual in I/O psychology. It is borrowed from the same playbook that designs licensing exams for highly regulated professions. What is unusual is applying that rigor to AI governance, where the field has been dominated by frameworks that were written, not validated.
See the framework in action
The fastest way to understand HAIL is to take the diagnostic. Fifteen minutes. Fifteen questions. A personalized maturity report grounded in the same five-domain structure that runs through the full certification.