NIST TEVV-Athlon
Source
NIST AI 200-2 ipd: The TEVV-Athlon Framework for Evaluating AI Systems
The TEVV-Athlon Framework provides a flexible method for constructing AI assessments around organizational evaluation objectives.
Its assessment structure uses measurement Blocks, Events, and Tools to collect evidence relevant to the characteristics being evaluated.
Pilot Investigation
The initial Structural Assurability investigation uses NIST's illustrative query-violation assessment in Section 3.
The research examines relationships among evaluative questions, available observations, evidentiary capabilities, evaluator conditions, and assumptions relevant to claim resolution.
Two initial finite-model constructions examine conditional claim-resolution questions. Their claims and modeling assumptions are researcher-defined not defined by NIST.
The investigation also considers what additional evidence or justification may be needed to connect sampled measurements to broader evaluative claims.
Research Status
TEVV-001 remains OPEN.
The initial experiment produces reproducible conditional results within its specified finite models.