Skip to content

R0021/2026-03-25/Q004 — ACH Matrix

Matrix

H1: Rigorous requirements H2: Minimal/vague H3: Adapting, incomplete
SRC01-E01: FAA roadmap, "methods must be developed" + -- ++
SRC02-E01: FDA GMLP 10 principles + change control ++ -- +
SRC03-E01: SR 11-7 "may lose effectiveness" for adaptive AI + -- ++
SRC04-E01: NIST voluntary RMF with TEVV + -- +

Legend: ++ Strongly supports · + Supports · -- Strongly contradicts · - Contradicts · N/A Not applicable

Diagnosticity Analysis

Most Diagnostic Evidence

Evidence ID Why Diagnostic
SRC03-E01 SR 11-7's explicit acknowledgment that validation approaches "may lose effectiveness" for AI is highly diagnostic — it confirms frameworks exist (contradicting H2) while revealing their limitations (supporting H3 over H1).

Least Diagnostic Evidence

Evidence ID Why Non-Diagnostic
SRC04-E01 NIST's voluntary framework is non-diagnostic because its voluntary nature makes it equally consistent with H1 (framework exists) and H3 (incomplete).

Outcome

Hypothesis supported: H3 — Frameworks exist but are adapting, with acknowledged gaps for modern AI systems.

Hypotheses eliminated: H2 — Multiple regulatory frameworks are in active use.

Hypotheses inconclusive: H1 — Partially supported; frameworks are rigorous but not yet fully adapted to AI.