Operational Integrity for Operational AI: A Technology-Neutral Benchmark Framework for Evaluating Operational AI Systems - Conceptual Specification Paper
Year: 2026
Type: article
Abstract: Artificial intelligence is increasingly moving from conversational assistance toward operational participation in enterprise systems, where AI components can invoke tools, execute workflows, modify data and create legally, technically or economically consequential outcomes. This paper introduces the Operational Integrity Benchmark (OIB), a technology-neutral framework for evaluating whether operational consequences arise through authorized, verified, governed, auditable and reproducible operational processes. The framework introduces Operational State Reconstruction, Canonical and Observed Operational States, Integrity Predicates, Operational Validation, benchmark scenarios, benchmark profiles, standardized metrics, an execution protocol and formal conformance requirements. The OIB is designed to enable reproducible and implementation-independent evaluation of heterogeneous operational AI systems without prescribing specific models, governance mechanisms or architectures.
Author: Gessler, Thomas
Language: English
Subfield: Safety Research
Field: Social Sciences
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W7203872880
FWCI:
Cites:
Cited by:
Related to: