• workshop_table17#tag-band S7 · 120–145 min

The academic theory contribution is a mechanism-based mid-range design theory of run-level governance for generative AI in organisational work. RAIDT does not present only a checklist, maturity model, or isolated empirical result. It contributes a designable governance object composed of a run-level evidence pack and a score profile, and it explains how governance artefacts plus influence methods as governance interventions generate observable outcomes under stated boundary conditions. That is the theory contribution: a bounded explanatory design theory that makes governance readiness inspectable, comparable, and testable.

It is Information Systems governance because the core question is how organisations operationalise accountability through information artefacts, routines, controls, and review practices. RAIDT focuses on configured use in context, not model capability in isolation. The issue is whether an organisation can later show what prompt, retrieval context, model configuration, tool use, oversight decision, and provenance were involved in one material use. That places the contribution squarely within Information Systems governance: it concerns information artefacts, governance mechanisms, record-keeping, auditability, and organisational control over socio-technical systems. The move to run as the unit of governance is especially important here, because it shifts governance from abstract policy statements to inspectable use events. In that sense, RAIDT extends Information Systems governance literature by identifying a distinct micro-operational governance problem for generative AI and by supplying the artefact logic needed to study it cumulatively.

Consider a healthcare note-summarisation service used in a hospital trust. The governance issue is not only whether the model summarises well, but whether the organisation can evidence how one summary was produced before it enters the patient record. A run-level evidence pack records the prompt template, model version, retrieval snapshot of guidance, uncertainty wording, clinician review, and the resulting score profile.

This is why the contribution belongs to Information Systems governance. The practical problem is one of organisational control over an information artefact embedded in workflow: who approves the template, which guidance corpus is authoritative, how review is logged, when a low Auditability or Traceability score blocks release, and how repeated runs are monitored for drift. Those are governance questions about routines, accountability, and evidence in use, rather than only technical questions about model architecture.

• 11-RAIDT_Academic_Logic_M_v11
• 12-RAIDT_DSR_Theory_M_v8