• integrated_82#Q2.10

RAIDT relates to socio-technical governance by operationalising it. The socio-technical papers argue that governance in digital systems is enacted through the interaction of artefacts, routines, roles, and decision rights rather than through technical controls or managerial policy alone. RAIDT translates that proposition into a governance method. By defining the run as the unit of governance, it creates a bounded object, the run-level evidence pack, that can travel across compliance, audit, operational, and managerial settings. The associated score profile expresses governance readiness in a structured way rather than leaving review to recollection or broad assurance claims.

RAIDT also gives a disciplined account of human-AI collaboration. In organisational GenAI use, collaboration is not a soft background condition; it is the mechanism through which prompts are set, sources are selected, outputs are reviewed, and reliance decisions are made. RAIDT captures these relations explicitly, so that human-AI collaboration becomes inspectable rather than assumed. This is where influence methods as governance interventions become central: structured prompting, retrieval augmentation, PEFT/LoRA, and preference-based alignment shape output behaviour, but they also shape what evidence exists for later review. Through the five pillars (Responsibility, Auditability, Interpretability, Dependability, Traceability), and the practical shorthand of anchors 1=missing / 3=partial / 5=audit-ready, RAIDT turns socio-technical governance into reviewable practice. It complements higher-level governance instruments by supplying the evidence layer through which collaboration in one run can be reconstructed, compared, and challenged.

In a cybersecurity alert-triage workflow, an analyst asks GenAI to recommend next steps on a suspicious login event. The assistant uses a structured prompt, a threat-intelligence retrieval layer, and policy constraints designed to discourage invented indicators. The analyst then decides whether to escalate the alert, close it, or seek more evidence. This is both socio-technical governance and human-AI collaboration in action: technical configuration, human judgement, and organisational control are inseparable.

RAIDT makes that collaboration governable by preserving the prompt, retrieval snapshot, tool outputs, model settings, variance checks from repeat runs, and the analyst's escalation note in one run-level evidence pack. A shared score profile lets operations, compliance, and audit inspect the same run without relying on memory. The organisation can then see whether the collaboration supported dependable, traceable, and auditable action, or whether the run needs challenge and redesign.

• 17-RAIDT-Sociotechnical_M_v6
• 11-RAIDT_Academic_Logic_M_v11
• 12-RAIDT_DSR_Theory_M_v8