• integrated_82#Q2.9

RAIDT relates to Information Systems governance as a complementary and operational extension rather than a replacement. The papers argue that IS governance already provides the right conceptual foundations: governance depends on artefacts, routines, roles, and decision rights organised so that digital action becomes reviewable and coordinated. RAIDT enters at the point where those established governance expectations become difficult to enact for generative AI, namely the configured use event. High-level policies, model cards, audit programmes, and procurement controls remain necessary, but they do not by themselves preserve one disputed run in a form that can be reconstructed across functions. RAIDT addresses that missing layer by defining the run as the unit of governance and by making a run-level evidence pack the proof object through which governance can be inspected in practice.

The relationship is therefore translational. RAIDT connects organisation-level governance intentions to use-level evidence, so that review, challenge, sampling, escalation, and learning can occur on the basis of preserved records rather than recollection. Its score profile gives a shared language for judging the five pillars (Responsibility, Auditability, Interpretability, Dependability, Traceability), and the anchors 1=missing / 3=partial / 5=audit-ready support comparable assessment across runs, teams, and suppliers. Crucially, the academic-logic paper shows that influence methods as governance interventions must also be governed, because retrieval, prompting, adaptation, and alignment change both outputs and evidentiary obligations. RAIDT therefore relates to IS governance by making configured GenAI use governable in the same socio-technical sense that IS governance has long required: not only controlled in principle, but reviewable through concrete information objects.

Consider cybersecurity alert triage. An organisation may already have an approved AI service, an incident policy, and managerial accountability for the workflow. RAIDT adds the missing IS-governance layer at the level of one alert triage run. The run-level evidence pack records the structured prompt, retrieval snapshot from the approved threat-intelligence source, model and tool versions, confidence markers, repeat-run stability checks, and any escalation to human supervision.

This lets the security lead, audit team, and compliance staff work from the same governance object. If the output is challenged after an incident, they can test Dependability and Auditability from evidence rather than from narrative assurance. The workflow is still governed by broader policy, but RAIDT makes that policy inspectable in one concrete case and shows whether the configuration was actually governance-ready.

• 11-RAIDT_Academic_Logic_M_v11
• 17-RAIDT-Sociotechnical_M_v6